3D Object Detection Evaluation 2017


The 3D object detection benchmark consists of 7481 training images and 7518 test images as well as the corresponding point clouds, comprising a total of 80.256 labeled objects. For evaluation, we compute precision-recall curves. To rank the methods we compute average precision. We require that all methods use the same parameter set for all test pairs. Our development kit provides details about the data format as well as MATLAB / C++ utility functions for reading and writing the label files.

We evaluate 3D object detection performance using the PASCAL criteria also used for 2D object detection. Far objects are thus filtered based on their bounding box height in the image plane. As only objects also appearing on the image plane are labeled, objects in don't car areas do not count as false positives. We note that the evaluation does not take care of ignoring detections that are not visible on the image plane — these detections might give rise to false positives. For cars we require an 3D bounding box overlap of 70%, while for pedestrians and cyclists we require a 3D bounding box overlap of 50%. Difficulties are defined as follows:

  • Easy: Min. bounding box height: 40 Px, Max. occlusion level: Fully visible, Max. truncation: 15 %
  • Moderate: Min. bounding box height: 25 Px, Max. occlusion level: Partly occluded, Max. truncation: 30 %
  • Hard: Min. bounding box height: 25 Px, Max. occlusion level: Difficult to see, Max. truncation: 50 %

All methods are ranked based on the moderately difficult results.

Note 2: On 08.10.2019, we have followed the suggestions of the Mapillary team in their paper Disentangling Monocular 3D Object Detection and use 40 recall positions instead of the 11 recall positions proposed in the original Pascal VOC benchmark. This results in a more fair comparison of the results, please check their paper. The last leaderboards right before this change can be found here: Object Detection Evaluation, 3D Object Detection Evaluation, Bird's Eye View Evaluation.
Important Policy Update: As more and more non-published work and re-implementations of existing work is submitted to KITTI, we have established a new policy: from now on, only submissions with significant novelty that are leading to a peer-reviewed paper in a conference or journal are allowed. Minor modifications of existing algorithms or student research projects are not allowed. Such work must be evaluated on a split of the training set. To ensure that our policy is adopted, new users must detail their status, describe their work and specify the targeted venue during registration. Furthermore, we will regularly delete all entries that are 6 months old but are still anonymous or do not have a paper associated with them. For conferences, 6 month is enough to determine if a paper has been accepted and to add the bibliography information. For longer review cycles, you need to resubmit your results.
Additional information used by the methods
  • Stereo: Method uses left and right (stereo) images
  • Flow: Method uses optical flow (2 temporally adjacent images)
  • Multiview: Method uses more than 2 temporally adjacent images
  • Laser Points: Method uses point clouds from Velodyne laser scanner
  • Additional training data: Use of additional data sources for training (see details)

Car


Method Setting Code Moderate Easy Hard Runtime Environment
1 SFD 84.76 % 91.73 % 77.92 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
2 Anonymous 83.96 % 90.83 % 77.47 % n/a s 1 core @ 2.5 Ghz (Python + C/C++)
3 DGDNH 83.88 % 90.69 % 79.50 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
4 GraR-VoI 83.27 % 91.89 % 77.78 % 0.07 s 1 core @ 2.5 Ghz (Python + C/C++)
5 VPFNet 83.21 % 91.02 % 78.20 % 0.06 s 2 cores @ 2.5 Ghz (Python)
H. Zhu, J. Deng, Y. Zhang, J. Ji, Q. Mao, H. Li and Y. Zhang: VPFNet: Improving 3D Object Detection with Virtual Point based LiDAR and Stereo Data Fusion. 2021.
6 GraR-Po 83.18 % 91.79 % 77.98 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
7 CasA 83.06 % 91.58 % 80.08 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
8 Anonymous 82.99 % 91.64 % 78.02 % 0.1 s GPU @ 2.5 Ghz (Python)
9 BtcDet
This method makes use of Velodyne laser scans.
 code 82.86 % 90.64 % 78.09 % 0.09 s GPU @ 2.5 Ghz (Python + C/C++)
Q. Xu, Y. Zhong and U. Neumann: Behind the Curtain: Learning Occluded Shapes for 3D Object Detection. Proceedings of the AAAI Conference on Artificial Intelligence 2022.
10 Anonymous 82.79 % 91.30 % 78.07 % n/a s 1 core @ 2.5 Ghz (C/C++)
11 GraR-Vo 82.77 % 91.29 % 77.20 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
12 PE-RCVN 82.69 % 91.51 % 77.75 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
13 CAD 82.68 % 88.96 % 77.91 % 0.1 s GPU @ 2.5 Ghz (Python + C/C++)
14 SPG_mini
This method makes use of Velodyne laser scans.
 code 82.66 % 90.64 % 77.91 % 0.09 s GPU @ 2.5 Ghz (Python)
Q. Xu, Y. Zhou, W. Wang, C. Qi and D. Anguelov: SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation. Proceedings of the IEEE conference on computer vision and pattern recognition (ICCV) 2021.
15 DSASNet 82.63 % 89.48 % 77.94 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
16 SA3DNet
This method uses stereo information.
This method makes use of Velodyne laser scans.
 82.57 % 90.49 % 77.88 % 0.05 s GPU @ 2.5 Ghz (Python)
17 SE-SSD
This method makes use of Velodyne laser scans.
 code 82.54 % 91.49 % 77.15 % 0.03 s 1 core @ 2.5 Ghz (Python + C/C++)
W. Zheng, W. Tang, L. Jiang and C. Fu: SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloud. CVPR 2021.
18 TF3D
This method makes use of Velodyne laser scans.
 82.46 % 89.10 % 77.78 % 0.1 s 2 cores @ 3.0 Ghz (Python)
19 DVF-V 82.45 % 89.40 % 77.56 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
20 GraR-Pi 82.42 % 90.94 % 77.00 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
21 DVF-PV 82.40 % 90.99 % 77.37 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
22 EA-M-RCNN(BorderAtt) 82.33 % 87.77 % 77.37 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
23 Anonymous 82.30 % 90.88 % 76.89 % 0.03 s 1 core @ 2.5 Ghz (Python + C/C++)
24 Anonymous code 82.28 % 90.55 % 77.59 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
25 CLOCs code 82.28 % 89.16 % 77.23 % 0.1 s 1 core @ 2.5 Ghz (Python)
S. Pang, D. Morris and H. Radha: CLOCs: Camera-LiDAR Object Candidates Fusion for 3D Object Detection . 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2020.
26 TBD
This method makes use of Velodyne laser scans.
 82.23 % 88.76 % 77.48 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
27 CityBrainLab 82.19 % 90.51 % 77.17 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
28 SPG
This method makes use of Velodyne laser scans.
 code 82.13 % 90.50 % 78.90 % 0.09 s 1 core @ 2.5 Ghz (Python + C/C++)
Q. Xu, Y. Zhou, W. Wang, C. Qi and D. Anguelov: SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation. Proceedings of the IEEE conference on computer vision and pattern recognition (ICCV) 2021.
29 VoTr-TSD code 82.09 % 89.90 % 79.14 % 0.07 s 1 core @ 2.5 Ghz (C/C++)
J. Mao, Y. Xue, M. Niu, H. Bai, J. Feng, X. Liang, H. Xu and C. Xu: Voxel Transformer for 3D Object Detection. ICCV 2021.
30 TBD 82.09 % 89.50 % 79.29 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
31 Pyramid R-CNN 82.08 % 88.39 % 77.49 % 0.07 s 1 core @ 2.5 Ghz (C/C++)
J. Mao, M. Niu, H. Bai, X. Liang, H. Xu and C. Xu: Pyramid R-CNN: Towards Better Performance and Adaptability for 3D Object Detection. ICCV 2021.
32 FS-Net
This method makes use of Velodyne laser scans.
 82.07 % 88.68 % 77.42 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
33 VoxSeT 82.06 % 88.53 % 77.46 % 0.09 s 1 core @ 2.5 Ghz (C/C++)
34 SRIF-RCNN 82.04 % 88.45 % 77.54 % 0.0947 s 1 core @ 2.5 Ghz (C/C++)
35 LGNet 82.02 % 90.65 % 77.34 % 0.07 s 1 core @ 2.5 Ghz (C/C++)
36 EQ-PVRCNN 82.01 % 90.13 % 77.53 % 0.2 s GPU @ 2.5 Ghz (Python + C/C++)
37 anonymous 81.99 % 88.82 % 77.26 % 0.09 s GPU @ 2.5 Ghz (Python)
38 Anonymous 81.96 % 89.90 % 77.20 % 0.1s 1 core @ 2.5 Ghz (C/C++)
39 EPNet++ 81.96 % 91.37 % 76.71 % 0.1 s GPU @ 2.5 Ghz (Python)
40 PV-RCNN++ 81.88 % 90.14 % 77.15 % 0.06 s 1 core @ 2.5 Ghz (C/C++)
41 PDV 81.86 % 90.43 % 77.36 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
42 SGNet 81.85 % 88.83 % 77.47 % 0.09 s GPU @ 2.5 Ghz (Python)
43 Anonymous 81.85 % 89.96 % 76.51 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
44 ST-RCNN
This method makes use of Velodyne laser scans.
 81.84 % 90.50 % 77.22 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
45 ST-RCNN (SNLW-RCNN)
This method makes use of Velodyne laser scans.
 code 81.84 % 90.50 % 77.22 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
46 ISE-RCNN 81.83 % 89.12 % 77.29 % 0.09 s 1 core @ 2.5 Ghz (Python + C/C++)
47 SqueezeRCNN 81.80 % 88.72 % 77.10 % 0.08 s 1 core @ 2.5 Ghz (Python)
48 CityBrainLab-CT3D code 81.77 % 87.83 % 77.16 % 0.07 s 1 core @ 2.5 Ghz (Python + C/C++)
H. Sheng, S. Cai, Y. Liu, B. Deng, J. Huang, X. Hua and M. Zhao: Improving 3D Object Detection with Channel- wise Transformer. ICCV 2021.
49 USVLab BSAODet (MM) 81.74 % 88.89 % 77.14 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
50 JPVNet 81.73 % 88.66 % 76.94 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
51 TBD 81.73 % 89.48 % 79.05 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
52 M3DeTR code 81.73 % 90.28 % 76.96 % n/a s GPU @ 1.0 Ghz (Python)
T. Guan, J. Wang, S. Lan, R. Chandra, Z. Wu, L. Davis and D. Manocha: M3DeTR: Multi-representation, Multi- scale, Mutual-relation 3D Object Detection with Transformers. 2021.
53 SIENet code 81.71 % 88.22 % 77.22 % 0.08 s 1 core @ 2.5 Ghz (Python)
Z. Li, Y. Yao, Z. Quan, W. Yang and J. Xie: SIENet: Spatial Information Enhancement Network for 3D Object Detection from Point Cloud. 2021.
54 TBD 81.71 % 88.46 % 76.63 % 0.1 s 1 core @ 2.5 Ghz (Python)
55 DCAnet 81.70 % 88.42 % 77.18 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
56 VCRCNN 81.68 % 90.52 % 77.26 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
57 TBD 81.68 % 87.93 % 76.92 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
58 ASCNet 81.67 % 88.48 % 76.93 % 0.09 s 1 core @ 2.5 Ghz (C/C++)
59 Fast VP-RCNN code 81.62 % 90.97 % 76.90 % 0.05 s 1 core @ 3.5 Ghz (C/C++)
60 Voxel R-CNN code 81.62 % 90.90 % 77.06 % 0.04 s GPU @ 3.0 Ghz (C/C++)
J. Deng, S. Shi, P. Li, W. Zhou, Y. Zhang and H. Li: Voxel R-CNN: Towards High Performance Voxel-based 3D Object Detection . AAAI 2021.
61 BADet code 81.61 % 89.28 % 76.58 % 0.11 s 1 core @ 2.5 Ghz (C/C++)
R. Qian, X. Lai and X. Li: BADet: Boundary-Aware 3D Object Detection from Point Clouds. Pattern Recognition (PR) 2022.
62 SARFE 81.59 % 88.88 % 76.74 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
63 HyBrid Feature Det 81.59 % 88.77 % 76.92 % 0.08 s 1 core @ 2.5 Ghz (Python + C/C++)
64 FromVoxelToPoint code 81.58 % 88.53 % 77.37 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
J. Li, H. Dai, L. Shao and Y. Ding: From Voxel to Point: IoU-guided 3D Object Detection for Point Cloud with Voxel-to- Point Decoder. MM '21: The 29th ACM International Conference on Multimedia (ACM MM) 2021.
65 H^23D R-CNN code 81.55 % 90.43 % 77.22 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
J. Deng, W. Zhou, Y. Zhang and H. Li: From Multi-View to Hollow-3D: Hallucinated Hollow-3D R-CNN for 3D Object Detection. IEEE Transactions on Circuits and Systems for Video Technology 2021.
66 anonymous code 81.55 % 90.94 % 76.74 % 0.05s 1 core @ >3.5 Ghz (python)
67 LZY_RCNN 81.52 % 88.77 % 78.59 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
68 MSG-PGNN 81.50 % 88.70 % 76.88 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
69 PTA-RCNN 81.48 % 88.18 % 76.98 % 0.08 s 1 core @ 2.5 Ghz (Python)
70 TransCyclistNet 81.46 % 88.47 % 76.87 % 0.08 s 1 core @ 2.5 Ghz (Python)
71 DSA-PV-RCNN
This method makes use of Velodyne laser scans.
 code 81.46 % 88.25 % 76.96 % 0.08 s 1 core @ 2.5 Ghz (Python + C/C++)
P. Bhattacharyya, C. Huang and K. Czarnecki: SA-Det3D: Self-Attention Based Context-Aware 3D Object Detection. 2021.
72 P2V-RCNN 81.45 % 88.34 % 77.20 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
J. Li, S. Luo, Z. Zhu, H. Dai, A. Krylov, Y. Ding and L. Shao: P2V-RCNN: Point to Voxel Feature Learning for 3D Object Detection from Point Clouds. IEEE Access 2021.
73 MMLab PV-RCNN
This method makes use of Velodyne laser scans.
 code 81.43 % 90.25 % 76.82 % 0.08 s 1 core @ 2.5 Ghz (Python + C/C++)
S. Shi, C. Guo, L. Jiang, Z. Wang, J. Shi, X. Wang and H. Li: PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detection. CVPR 2020.
74 TPCG 81.41 % 89.16 % 76.90 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
75 DDet 81.38 % 89.63 % 78.57 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
76 WHUT-iou_ssd code 81.37 % 89.84 % 76.83 % 0.045s 1 core @ 2.5 Ghz (C/C++)
77 XView 81.35 % 89.21 % 76.87 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
L. Xie, G. Xu, D. Cai and X. He: X-view: Non-egocentric Multi-View 3D Object Detector. 2021.
78 ISE-RCNN-PV 81.34 % 88.05 % 76.99 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
79 RangeRCNN
This method makes use of Velodyne laser scans.
 81.33 % 88.47 % 77.09 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Liang, M. Zhang, Z. Zhang, X. Zhao and S. Pu: RangeRCNN: Towards Fast and Accurate 3D Object Detection with Range Image Representation. arXiv preprint arXiv:2009.00206 2020.
80 CAT-Det 81.32 % 89.87 % 76.68 % 0.3 s GPU @ 2.5 Ghz (Python + C/C++)
81 TransDet3D 81.28 % 88.11 % 76.73 % 0.08 s 1 core @ 2.5 Ghz (Python)
82 Generalized-SIENet 81.24 % 87.70 % 76.79 % 0.08 s 1 core @ 2.5 Ghz (Python)
83 Point Image Fusion 81.23 % 89.01 % 76.77 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
84 SAA-PV-RCNN 81.09 % 87.24 % 78.05 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
85 FPC-RCNN 81.08 % 88.68 % 76.46 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
86 VPFNet code 80.97 % 88.51 % 76.74 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
C. Wang, H. Chen and L. Fu: VPFNet: Voxel-Pixel Fusion Network for Multi-class 3D Object Detection. 2021.
87 CSVoxel-RCNN 80.97 % 87.66 % 76.29 % 0.03 s GPU @ 1.0 Ghz (Python)
88 VueronNet code 80.96 % 90.06 % 73.72 % 0.06 s 1 core @ 2.0 Ghz (Python)
89 FusionDetv2-v4 80.93 % 87.75 % 76.12 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
90 AIMC-RUC 80.83 % 90.14 % 73.59 % 0.11 s 1 core @ 2.5 Ghz (Python)
91 sa-voxel-centernet code 80.77 % 87.39 % 76.45 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
92 SA-voxel-centernet code 80.77 % 87.28 % 76.51 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
93 Associate-3Ddet_v2 80.77 % 91.53 % 75.23 % 0.04 s 1 core @ 2.5 Ghz (Python)
94 FusionDetv2-v3 80.70 % 88.05 % 76.10 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
95 StructuralIF 80.69 % 87.15 % 76.26 % 0.02 s 8 cores @ 2.5 Ghz (Python)
J. Pei An: Deep structural information fusion for 3D object detection on LiDAR-camera system. Accepted in CVIU 2021.
96 SPVB-SSD 80.68 % 86.99 % 76.23 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
97 CLOCs_PVCas code 80.67 % 88.94 % 77.15 % 0.1 s 1 core @ 2.5 Ghz (Python)
S. Pang, D. Morris and H. Radha: CLOCs: Camera-LiDAR Object Candidates Fusion for 3D Object Detection . 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2020.
98 SCIR-Net
This method makes use of Velodyne laser scans.
 80.62 % 87.53 % 76.00 % 0.03 s GPU @ 2.5 Ghz (Python + C/C++)
99 GVNet code 80.52 % 87.63 % 75.99 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
100 SVGA-Net 80.47 % 87.33 % 75.91 % 0.03s 1 core @ 2.5 Ghz (Python + C/C++)
Q. He, Z. Wang, H. Zeng, Y. Zeng and Y. Liu: SVGA-Net: Sparse Voxel-Graph Attention Network for 3D Object Detection from Point Clouds. AAAI 2022.
101 TBD 80.44 % 88.83 % 73.18 % 0.1 s 1 core @ 2.5 Ghz (Python)
102 Sem-Aug v1 code 80.40 % 88.92 % 77.37 % 0.04 s GPU @ 3.5 Ghz (Python)
103 SRDL 80.38 % 87.73 % 76.27 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
ERROR: Wrong syntax in BIBTEX file.
104 Fast-CLOCs 80.35 % 89.10 % 76.99 % 0.1 s GPU @ 2.5 Ghz (Python)
S. Pang, D. Morris and H. Radha: Fast-CLOCs: Fast Camera-LiDAR Object Candidates Fusion for 3D Object Detection. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2022.
105 FPV-SSD 80.34 % 87.72 % 75.40 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
106 SPANet 80.34 % 91.05 % 74.89 % 0.06 s 1 core @ 2.5 Ghz (C/C++)
Y. Ye: SPANet: Spatial and Part-Aware Aggregation Network for 3D Object Detection. Pacific Rim International Conference on Artificial Intelligence 2021.
107 IA-SSD (single) 80.32 % 88.87 % 75.10 % 0.013 s 1 core @ 2.5 Ghz (C/C++)
108 AM-SSD 80.30 % 89.58 % 75.02 % 0.04 s 1 core @ 2.5 Ghz (Python)
109 TBD 80.29 % 87.37 % 73.05 % 0.1 s 1 core @ 2.5 Ghz (Python)
110 CIA-SSD
This method makes use of Velodyne laser scans.
 code 80.28 % 89.59 % 72.87 % 0.03 s 1 core @ 2.5 Ghz (Python + C/C++)
W. Zheng, W. Tang, S. Chen, L. Jiang and C. Fu: CIA-SSD: Confident IoU-Aware Single-Stage Object Detector From Point Cloud. AAAI 2021.
111 FusionDetv1 80.28 % 87.45 % 76.21 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
112 DVF 80.21 % 88.97 % 75.22 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
113 VCT 80.19 % 89.12 % 77.19 % 0.2 s 1 core @ 2.5 Ghz (Python)
114 IA-SSD (multi) 80.13 % 88.34 % 75.04 % 0.014 s 1 core @ 2.5 Ghz (C/C++)
115 TBD 80.12 % 88.30 % 75.29 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
116 EBM3DOD code 80.12 % 91.05 % 72.78 % 0.12 s 1 core @ 2.5 Ghz (Python)
F. Gustafsson, M. Danelljan and T. Schön: Accurate 3D Object Detection using Energy- Based Models. arXiv preprint arXiv:2012.04634 2020.
117 TBD 80.12 % 86.50 % 75.72 % 0.06 s GPU @ 2.5 Ghz (Python)
118 3D-CVF at SPA
This method makes use of Velodyne laser scans.
 80.05 % 89.20 % 73.11 % 0.06 s 1 core @ 2.5 Ghz (C/C++)
J. Yoo, Y. Kim, J. Kim and J. Choi: 3D-CVF: Generating Joint Camera and LiDAR Features Using Cross-View Spatial Feature Fusion for 3D Object Detection. ECCV 2020.
119 TBD 80.02 % 88.45 % 74.85 % TBD GPU @ 2.5 Ghz (Python + C/C++)
120 MVOD 80.01 % 88.53 % 77.24 % 0.16 s 1 core @ 2.5 Ghz (C/C++)
121 MBDF-Net 80.00 % 90.87 % 75.04 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
122 SIF 79.88 % 86.84 % 75.89 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
P. An: SIF. Submitted to CVIU 2021.
123 RangeIoUDet
This method makes use of Velodyne laser scans.
 79.80 % 88.60 % 76.76 % 0.02 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Liang, Z. Zhang, M. Zhang, X. Zhao and S. Pu: RangeIoUDet: Range Image Based Real-Time 3D Object Detector Optimized by Intersection Over Union. CVPR 2021.
124 SA-SSD code 79.79 % 88.75 % 74.16 % 0.04 s 1 core @ 2.5 Ghz (Python)
C. He, H. Zeng, J. Huang, X. Hua and L. Zhang: Structure Aware Single-stage 3D Object Detection from Point Cloud. CVPR 2020.
125 DGT-Det3D 79.78 % 86.76 % 75.73 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
126 KpNet 79.75 % 88.92 % 72.17 % 0.42 s 1 core @ 2.5 Ghz (C/C++)
127 KpNet 79.74 % 88.88 % 72.13 % 42 s 1 core @ 2.5 Ghz (C/C++)
128 STD code 79.71 % 87.95 % 75.09 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Yang, Y. Sun, S. Liu, X. Shen and J. Jia: STD: Sparse-to-Dense 3D Object Detector for Point Cloud. ICCV 2019.
129 MGAF-3DSSD code 79.68 % 88.16 % 72.39 % 0.1 s 1 core @ 2.5 Ghz (Python)
J. Li, H. Dai, L. Shao and Y. Ding: Anchor-free 3D Single Stage Detector with Mask-Guided Attention for Point Cloud. MM '21: The 29th ACM International Conference on Multimedia (ACM MM) 2021.
130 MBDF-Net-1 79.65 % 90.43 % 74.72 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
131 Struc info fusion II 79.59 % 88.97 % 72.51 % 0.05 s GPU @ 2.5 Ghz (Python)
P. An, J. Liang, J. Ma, K. Yu and B. Fang: Struc info fusion. Submitted to CVIU 2021.
132 3DSSD code 79.57 % 88.36 % 74.55 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Yang, Y. Sun, S. Liu and J. Jia: 3DSSD: Point-based 3D Single Stage Object Detector. CVPR 2020.
133 EBM3DOD baseline code 79.52 % 88.80 % 72.30 % 0.05 s 1 core @ 2.5 Ghz (Python)
F. Gustafsson, M. Danelljan and T. Schön: Accurate 3D Object Detection using Energy- Based Models. arXiv preprint arXiv:2012.04634 2020.
134 Struc info fusion I 79.49 % 88.70 % 74.25 % 0.05 s 1 core @ 2.5 Ghz (Python)
P. An, J. Liang, J. Ma, K. Yu and B. Fang: Struc info fusion. Submitted to CVIU 2021.
135 Point-GNN
This method makes use of Velodyne laser scans.
 code 79.47 % 88.33 % 72.29 % 0.6 s GPU @ 2.5 Ghz (Python)
W. Shi and R. Rajkumar: Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud. CVPR 2020.
136 SECOND 79.46 % 87.44 % 73.97 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
137 NV-RCNN 79.32 % 87.58 % 74.74 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
138 USVLab BSAODet (SM) 79.30 % 88.02 % 76.05 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
139 EPNet code 79.28 % 89.81 % 74.59 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
T. Huang, Z. Liu, X. Chen and X. Bai: EPNet: Enhancing Point Features with Image Semantics for 3D Object Detection. ECCV 2020.
140 FPCR-CNN 79.25 % 88.45 % 75.69 % 0.05 s 1 core @ 2.5 Ghz (Python)
141 DVFENet 79.18 % 86.20 % 74.58 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
Y. He, G. Xia, Y. Luo, L. Su, Z. Zhang, W. Li and P. Wang: DVFENet: Dual-branch Voxel Feature Extraction Network for 3D Object Detection. Neurocomputing 2021.
142 Faraway-Frustum
This method makes use of Velodyne laser scans.
 code 79.05 % 87.45 % 76.14 % 0.1 s GPU @ 2.5 Ghz (Python)
H. Zhang, D. Yang, E. Yurtsever, K. Redmill and U. Ozguner: Faraway-Frustum: Dealing with Lidar Sparsity for 3D Object Detection using Fusion. arXiv preprint arXiv:2011.01404 2020.
143 3D IoU-Net 79.03 % 87.96 % 72.78 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
J. Li, S. Luo, Z. Zhu, H. Dai, S. Krylov, Y. Ding and L. Shao: 3D IoU-Net: IoU Guided 3D Object Detector for Point Clouds. arXiv preprint arXiv:2004.04962 2020.
144 SERCNN
This method makes use of Velodyne laser scans.
 78.96 % 87.74 % 74.30 % 0.1 s 1 core @ 2.5 Ghz (Python)
D. Zhou, J. Fang, X. Song, L. Liu, J. Yin, Y. Dai, H. Li and R. Yang: Joint 3D Instance Segmentation and Object Detection for Autonomous Driving. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020.
145 NV2P-RCNN 78.92 % 87.36 % 74.16 % 0.1 s GPU @ 2.5 Ghz (Python)
146 demo 78.85 % 87.50 % 72.05 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
147 MSADet 78.81 % 88.31 % 73.82 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
148 FPC3D
This method makes use of the epipolar geometry.
 78.81 % 87.61 % 75.49 % 33 s 1 core @ 2.5 Ghz (C/C++)
149 MVAF-Net code 78.71 % 87.87 % 75.48 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
G. Wang, B. Tian, Y. Zhang, L. Chen, D. Cao and J. Wu: Multi-View Adaptive Fusion Network for 3D Object Detection. arXiv preprint arXiv:2011.00652 2020.
150 MMLab-PartA^2
This method makes use of Velodyne laser scans.
 code 78.49 % 87.81 % 73.51 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
S. Shi, Z. Wang, J. Shi, X. Wang and H. Li: From Points to Parts: 3D Object Detection from Point Cloud with Part-aware and Part-aggregation Network. IEEE Transactions on Pattern Analysis and Machine Intelligence 2020.
151 CLOCs_SecCas 78.45 % 86.38 % 72.45 % 0.1 s 1 core @ 2.5 Ghz (Python)
S. Pang, D. Morris and H. Radha: CLOCs: Camera-LiDAR Object Candidates Fusion for 3D Object Detection. 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2020.
152 CSNet 78.42 % 87.39 % 71.75 % 0.1 s 1 core @ 2.5 Ghz (Python)
153 FusionDetv2-v2 78.42 % 86.59 % 73.87 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
154 Patches - EMP
This method makes use of Velodyne laser scans.
 78.41 % 89.84 % 73.15 % 0.5 s GPU @ 2.5 Ghz (Python)
J. Lehner, A. Mitterecker, T. Adler, M. Hofmarcher, B. Nessler and S. Hochreiter: Patch Refinement: Localized 3D Object Detection. arXiv preprint arXiv:1910.04093 2019.
155 HotSpotNet 78.31 % 87.60 % 73.34 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
Q. Chen, L. Sun, Z. Wang, K. Jia and A. Yuille: object as hotspots. Proceedings of the European Conference on Computer Vision (ECCV) 2020.
156 FusionDetv2-v5 78.30 % 86.94 % 73.44 % 0.05 s 1 core @ 2.5 Ghz (Java + C/C++)
157 SAA-SECOND 78.13 % 86.13 % 73.34 % 38m s 1 core @ 2.5 Ghz (C/C++)
158 3D-VDNet 78.05 % 87.13 % 72.90 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
159 VPN 77.93 % 85.02 % 72.97 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
160 CenterNet3D 77.90 % 86.20 % 73.03 % 0.04 s GPU @ 1.5 Ghz (Python)
G. Wang, B. Tian, Y. Ai, T. Xu, L. Chen and D. Cao: CenterNet3D:An Anchor free Object Detector for Autonomous Driving. 2020.
161 TBD 77.85 % 86.46 % 72.39 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
162 CVFNet 77.70 % 88.75 % 71.95 % 28.1ms 1 core @ 2.5 Ghz (Python)
163 AutoAlign 77.58 % 86.84 % 73.23 % 0.1 s 1 core @ 2.5 Ghz (Python)
164 TBD 77.56 % 85.38 % 72.32 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
165 TCDVF 77.49 % 85.55 % 72.29 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
166 UberATG-MMF
This method makes use of Velodyne laser scans.
 77.43 % 88.40 % 70.22 % 0.08 s GPU @ 2.5 Ghz (Python)
M. Liang*, B. Yang*, Y. Chen, R. Hu and R. Urtasun: Multi-Task Multi-Sensor Fusion for 3D Object Detection. CVPR 2019.
167 Associate-3Ddet code 77.40 % 85.99 % 70.53 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
L. Du, X. Ye, X. Tan, J. Feng, Z. Xu, E. Ding and S. Wen: Associate-3Ddet: Perceptual-to-Conceptual Association for 3D Point Cloud Object Detection. The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2020.
168 Fast Point R-CNN
This method makes use of Velodyne laser scans.
 77.40 % 85.29 % 70.24 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Chen, S. Liu, X. Shen and J. Jia: Fast Point R-CNN. Proceedings of the IEEE international conference on computer vision (ICCV) 2019.
169 3D_att
This method makes use of Velodyne laser scans.
 77.27 % 88.46 % 70.11 % 0.17 s GPU @ 2.5 Ghz (Python)
170 Patches
This method makes use of Velodyne laser scans.
 77.20 % 88.67 % 71.82 % 0.15 s GPU @ 2.0 Ghz
J. Lehner, A. Mitterecker, T. Adler, M. Hofmarcher, B. Nessler and S. Hochreiter: Patch Refinement: Localized 3D Object Detection. arXiv preprint arXiv:1910.04093 2019.
171 Sem-Aug-PointRCNN code 77.04 % 82.75 % 73.21 % 0.1 s GPU @ 3.5 Ghz (C/C++)
172 HRI-VoxelFPN 76.70 % 85.64 % 69.44 % 0.02 s GPU @ 2.5 Ghz (Python + C/C++)
H. Kuang, B. Wang, J. An, M. Zhang and Z. Zhang: Voxel-FPN:multi-scale voxel feature aggregation in 3D object detection from point clouds. sensors 2020.
173 DKAnet 76.70 % 84.57 % 71.54 % 0.05 s 1 core @ 2.0 Ghz (Python)
174 SARPNET 76.64 % 85.63 % 71.31 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
Y. Ye, H. Chen, C. Zhang, X. Hao and Z. Zhang: SARPNET: Shape Attention Regional Proposal Network for LiDAR-based 3D Object Detection. Neurocomputing 2019.
175 YF 76.57 % 87.15 % 71.23 % 0.04 s GPU @ 2.5 Ghz (C/C++)
176 3D IoU Loss
This method makes use of Velodyne laser scans.
 76.50 % 86.16 % 71.39 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
D. Zhou, J. Fang, X. Song, C. Guan, J. Yin, Y. Dai and R. Yang: IoU Loss for 2D/3D Object Detection. International Conference on 3D Vision (3DV) 2019.
177 HNet-3DSSD
This method makes use of Velodyne laser scans.
 code 76.48 % 86.06 % 69.71 % 0.05 s GPU @ 2.5 Ghz (Python)
178 F-ConvNet
This method makes use of Velodyne laser scans.
 code 76.39 % 87.36 % 66.69 % 0.47 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Wang and K. Jia: Frustum ConvNet: Sliding Frustums to Aggregate Local Point-Wise Features for Amodal 3D Object Detection. IROS 2019.
179 SegVoxelNet 76.13 % 86.04 % 70.76 % 0.04 s 1 core @ 2.5 Ghz (Python)
H. Yi, S. Shi, M. Ding, J. Sun, K. Xu, H. Zhou, Z. Wang, S. Li and G. Wang: SegVoxelNet: Exploring Semantic Context and Depth-aware Features for 3D Vehicle Detection from Point Cloud. ICRA 2020.
180 S-AT GCN 76.04 % 83.20 % 71.17 % 0.02 s GPU @ 2.0 Ghz (Python)
L. Wang, C. Wang, X. Zhang, T. Lan and J. Li: S-AT GCN: Spatial-Attention Graph Convolution Network based Feature Enhancement for 3D Object Detection. CoRR 2021.
181 TANet code 75.94 % 84.39 % 68.82 % 0.035s GPU @ 2.5 Ghz (Python + C/C++)
Z. Liu, X. Zhao, T. Huang, R. Hu, Y. Zhou and X. Bai: TANet: Robust 3D Object Detection from Point Clouds with Triple Attention. AAAI 2020.
182 APL-Second 75.75 % 84.26 % 70.65 % 0.05 s 1 core @ 2.5 Ghz (Python)
183 PointRGCN 75.73 % 85.97 % 70.60 % 0.26 s GPU @ V100 (Python)
J. Zarzar, S. Giancola and B. Ghanem: PointRGCN: Graph Convolution Networks for 3D Vehicles Detection Refinement. ArXiv 2019.
184 MMLab-PointRCNN
This method makes use of Velodyne laser scans.
 code 75.64 % 86.96 % 70.70 % 0.1 s GPU @ 2.5 Ghz (Python + C/C++)
S. Shi, X. Wang and H. Li: Pointrcnn: 3d object proposal generation and detection from point cloud. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2019.
185 AB3DMOT
This method makes use of Velodyne laser scans.
This is an online method (no batch processing).
 code 75.43 % 86.10 % 68.88 % 0.0047s 1 core @ 2.5 Ghz (Python)
X. Weng and K. Kitani: A Baseline for 3D Multi-Object Tracking. arXiv:1907.03961 2019.
186 R-GCN 75.26 % 83.42 % 68.73 % 0.16 s GPU @ 2.5 Ghz (Python)
J. Zarzar, S. Giancola and B. Ghanem: PointRGCN: Graph Convolution Networks for 3D Vehicles Detection Refinement. ArXiv 2019.
187 epBRM
This method makes use of Velodyne laser scans.
 code 75.15 % 85.00 % 69.84 % 0.1 s GPU @ >3.5 Ghz (Python + C/C++)
K. Shin: Improving a Quality of 3D Object Detection by Spatial Transformation Mechanism. arXiv preprint arXiv:1910.04853 2019.
188 MAFF-Net(DAF-Pillar) 75.04 % 85.52 % 67.61 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
Z. Zhang, Z. Liang, M. Zhang, X. Zhao, Y. Ming, T. Wenming and S. Pu: MAFF-Net: Filter False Positive for 3D Vehicle Detection with Multi-modal Adaptive Feature Fusion. arXiv preprint arXiv:2009.10945 2020.
189 sscl-20p 74.82 % 86.06 % 69.87 % 0.02 s 1 core @ 2.5 Ghz (Python)
190 PI-RCNN 74.82 % 84.37 % 70.03 % 0.1 s 1 core @ 2.5 Ghz (Python)
L. Xie, C. Xiang, Z. Yu, G. Xu, Z. Yang, D. Cai and X. He: PI-RCNN: An Efficient Multi-sensor 3D Object Detector with Point-based Attentive Cont-conv Fusion Module. AAAI 2020 : The Thirty-Fourth AAAI Conference on Artificial Intelligence 2020.
191 MF 74.70 % 83.42 % 66.51 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
192 LazyTorch-CP-Infer-O 74.57 % 81.82 % 70.24 % 1 s 1 core @ 2.5 Ghz (C/C++)
193 FPC3D_all
This method makes use of Velodyne laser scans.
 74.55 % 85.50 % 69.91 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
194 LazyTorch-CP-Small-P 74.44 % 81.73 % 70.14 % 1 s 1 core @ 2.5 Ghz (C/C++)
195 City-CF-fixed 74.37 % 83.23 % 69.65 % 1 s 1 core @ 2.5 Ghz (C/C++)
196 PointPillars
This method makes use of Velodyne laser scans.
 code 74.31 % 82.58 % 68.99 % 16 ms 1080ti GPU and Intel i7 CPU
A. Lang, S. Vora, H. Caesar, L. Zhou, J. Yang and O. Beijbom: PointPillars: Fast Encoders for Object Detection from Point Clouds. CVPR 2019.
197 ARPNET 74.04 % 84.69 % 68.64 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Ye, C. Zhang and X. Hao: ARPNET: attention region proposal network for 3D object detection. Science China Information Sciences 2019.
198 LazyTorch-CP 73.96 % 81.17 % 69.48 % 1 s 1 core @ 2.5 Ghz (C/C++)
199 PC-CNN-V2
This method makes use of Velodyne laser scans.
 73.79 % 85.57 % 65.65 % 0.5 s GPU @ 2.5 Ghz (Matlab + C/C++)
X. Du, M. Ang, S. Karaman and D. Rus: A General Pipeline for 3D Detection of Vehicles. 2018 IEEE International Conference on Robotics and Automation (ICRA) 2018.
200 C-GCN 73.62 % 83.49 % 67.01 % 0.147 s GPU @ V100 (Python)
J. Zarzar, S. Giancola and B. Ghanem: PointRGCN: Graph Convolution Networks for 3D Vehicles Detection Refinement. ArXiv 2019.
201 3DBN
This method makes use of Velodyne laser scans.
 73.53 % 83.77 % 66.23 % 0.13s 1080Ti (Python+C/C++)
X. Li, J. Guivant, N. Kwok and Y. Xu: 3D Backbone Network for 3D Object Detection. CoRR 2019.
202 Dune-DCF-e11 73.51 % 80.89 % 68.89 % 1 s 1 core @ 2.5 Ghz (C/C++)
203 CrazyTensor-CP 73.50 % 81.04 % 69.87 % 1 s 1 core @ 2.5 Ghz (Python)
204 PointRGBNet 73.49 % 83.99 % 68.56 % 0.08 s 4 cores @ 2.5 Ghz (Python + C/C++)
205 City-CF 73.48 % 80.85 % 69.07 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
206 RangeDet code 73.44 % 80.53 % 67.28 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
207 Dune-DCF-e15 73.29 % 80.34 % 68.54 % 1 s 1 core @ 2.5 Ghz (C/C++)
208 SCNet
This method makes use of Velodyne laser scans.
 73.17 % 83.34 % 67.93 % 0.04 s GPU @ 3.0 Ghz (Python)
Z. Wang, H. Fu, L. Wang, L. Xiao and B. Dai: SCNet: Subdivision Coding Network for Object Detection Based on 3D Point Cloud. IEEE Access 2019.
209 Dune-DCF-e09 73.15 % 80.40 % 68.57 % 1 s 1 core @ 2.5 Ghz (C/C++)
210 PP-PCdet code 73.07 % 83.32 % 68.18 % 0.01 s 1 core @ 2.5 Ghz (Python)
211 PFF3D
This method makes use of Velodyne laser scans.
 code 72.93 % 81.11 % 67.24 % 0.05 s GPU @ 3.0 Ghz (Python + C/C++)
L. Wen and K. Jo: Fast and Accurate 3D Object Detection for Lidar-Camera-Based Autonomous Vehicles Using One Shared Voxel-Based Backbone. IEEE Access 2021.
212 CrazyTensor-CF 72.92 % 79.87 % 68.41 % 1 s 1 core @ 2.5 Ghz (C/C++)
213 DASS 72.31 % 81.85 % 65.99 % 0.09 s 1 core @ 2.0 Ghz (Python)
O. Unal, L. Van Gool and D. Dai: Improving Point Cloud Semantic Segmentation by Learning 3D Object Detection. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2021.
214 HS3D code 72.25 % 83.57 % 67.49 % 0.12 s 1 core @ 2.5 Ghz (Python + C/C++)
215 TBD_BD code 72.16 % 83.36 % 66.87 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
216 TBD 71.94 % 83.20 % 66.83 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
217 AVOD-FPN
This method makes use of Velodyne laser scans.
 code 71.76 % 83.07 % 65.73 % 0.1 s Titan X (Pascal)
J. Ku, M. Mozifian, J. Lee, A. Harakeh and S. Waslander: Joint 3D Proposal Generation and Object Detection from View Aggregation. IROS 2018.
218 PointPainting
This method makes use of Velodyne laser scans.
 71.70 % 82.11 % 67.08 % 0.4 s GPU @ 2.5 Ghz (Python + C/C++)
S. Vora, A. Lang, B. Helou and O. Beijbom: PointPainting: Sequential Fusion for 3D Object Detection. CVPR 2020.
219 Contrastive PP code 71.64 % 84.80 % 66.49 % 0.01 s 1 core @ 2.5 Ghz (Python)
220 WS3D
This method makes use of Velodyne laser scans.
 70.59 % 80.99 % 64.23 % 0.1 s GPU @ 2.5 Ghz (Python)
Q. Meng, W. Wang, T. Zhou, J. Shen, L. Van Gool and D. Dai: Weakly Supervised 3D Object Detection from Lidar Point Cloud. 2020.
221 F-PointNet
This method makes use of Velodyne laser scans.
 code 69.79 % 82.19 % 60.59 % 0.17 s GPU @ 3.0 Ghz (Python)
C. Qi, W. Liu, C. Wu, H. Su and L. Guibas: Frustum PointNets for 3D Object Detection from RGB-D Data. arXiv preprint arXiv:1711.08488 2017.
222 FusionDetv2-baseline 68.87 % 79.05 % 63.68 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
223 UberATG-ContFuse
This method makes use of Velodyne laser scans.
 68.78 % 83.68 % 61.67 % 0.06 s GPU @ 2.5 Ghz (Python)
M. Liang, B. Yang, S. Wang and R. Urtasun: Deep Continuous Fusion for Multi-Sensor 3D Object Detection. ECCV 2018.
224 MLOD
This method makes use of Velodyne laser scans.
 code 67.76 % 77.24 % 62.05 % 0.12 s GPU @ 1.5 Ghz (Python)
J. Deng and K. Czarnecki: MLOD: A multi-view 3D object detection based on robust feature fusion method. arXiv preprint arXiv:1909.04163 2019.
225 DSGN++
This method uses stereo information.
 67.37 % 83.21 % 59.91 % 0.4 s NVIDIA Tesla V100
226 DisposalNet
This method uses stereo information.
 67.33 % 77.55 % 62.44 % 0.2 s 1 core @ 2.5 Ghz (Python + C/C++)
ERROR: Wrong syntax in BIBTEX file.
227 AVOD
This method makes use of Velodyne laser scans.
 code 66.47 % 76.39 % 60.23 % 0.08 s Titan X (pascal)
J. Ku, M. Mozifian, J. Lee, A. Harakeh and S. Waslander: Joint 3D Proposal Generation and Object Detection from View Aggregation. IROS 2018.
228 FusionDetv2-v1 65.65 % 75.21 % 60.65 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
229 MMLAB LIGA-Stereo
This method uses stereo information.
 code 64.66 % 81.39 % 57.22 % 0.4 s 1 core @ 2.5 Ghz (Python + C/C++)
X. Guo, S. Shi, X. Wang and H. Li: LIGA-Stereo: Learning LiDAR Geometry Aware Representations for Stereo-based 3D Detector. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) 2021.
230 BirdNet+
This method makes use of Velodyne laser scans.
 code 64.04 % 76.15 % 59.79 % 0.11 s Titan Xp (PyTorch)
A. Barrera, J. Beltrán, C. Guindel, J. Iglesias and F. García: BirdNet+: Two-Stage 3D Object Detection in LiDAR through a Sparsity-Invariant Bird’s Eye View. IEEE Access 2021.
231 MV3D
This method makes use of Velodyne laser scans.
 63.63 % 74.97 % 54.00 % 0.36 s GPU @ 2.5 Ghz (Python + C/C++)
X. Chen, H. Ma, J. Wan, B. Li and T. Xia: Multi-View 3D Object Detection Network for Autonomous Driving. CVPR 2017.
232 KMC code 62.74 % 74.45 % 56.76 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
233 LIGA-Stereo-old
This method uses stereo information.
 62.65 % 81.76 % 55.24 % 0.375 s Titan Xp
234 SD3DOD 62.00 % 76.09 % 55.46 % 0.04 s GPU @ 2.5 Ghz (Python)
235 AEC3D 61.99 % 72.16 % 57.11 % 18 ms GPU @ 2.5 Ghz (Python)
236 VN3D 61.41 % 72.37 % 56.86 % 0.02 s 1 core @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
237 RCD 60.56 % 70.54 % 55.58 % 0.1 s GPU @ 2.5 Ghz (Python)
A. Bewley, P. Sun, T. Mensink, D. Anguelov and C. Sminchisescu: Range Conditioned Dilated Convolutions for Scale Invariant 3D Object Detection. Conference on Robot Learning (CoRL) 2020.
238 A3DODWTDA
This method makes use of Velodyne laser scans.
 code 56.82 % 62.84 % 48.12 % 0.08 s GPU @ 3.0 Ghz (Python)
F. Gustafsson and E. Linder-Norén: Automotive 3D Object Detection Without Target Domain Annotations. 2018.
239 PL++ (SDN+GDC)
This method uses stereo information.
This method makes use of Velodyne laser scans.
 code 54.88 % 68.38 % 49.16 % 0.6 s GPU @ 2.5 Ghz (C/C++)
Y. You, Y. Wang, W. Chao, D. Garg, G. Pleiss, B. Hariharan, M. Campbell and K. Weinberger: Pseudo-LiDAR++: Accurate Depth for 3D Object Detection in Autonomous Driving. International Conference on Learning Representations 2020.
240 MV3D (LIDAR)
This method makes use of Velodyne laser scans.
 54.54 % 68.35 % 49.16 % 0.24 s GPU @ 2.5 Ghz (Python + C/C++)
X. Chen, H. Ma, J. Wan, B. Li and T. Xia: Multi-View 3D Object Detection Network for Autonomous Driving. CVPR 2017.
241 CDN
This method uses stereo information.
 code 54.22 % 74.52 % 46.36 % 0.6 s GPU @ 2.5 Ghz (Python)
D. Garg, Y. Wang, B. Hariharan, M. Campbell, K. Weinberger and W. Chao: Wasserstein Distances for Stereo Disparity Estimation. Advances in Neural Information Processing Systems (NeurIPS) 2020.
242 CG-Stereo
This method uses stereo information.
 53.58 % 74.39 % 46.50 % 0.57 s GeForce RTX 2080 Ti
C. Li, J. Ku and S. Waslander: Confidence Guided Stereo 3D Object Detection with Split Depth Estimation. IROS 2020.
243 UPF_3D
This method uses stereo information.
 52.83 % 78.24 % 46.12 % 0.29 s 1 core @ 2.5 Ghz (Python)
244 DSGN
This method uses stereo information.
 code 52.18 % 73.50 % 45.14 % 0.67 s NVIDIA Tesla V100
Y. Chen, S. Liu, X. Shen and J. Jia: DSGN: Deep Stereo Geometry Network for 3D Object Detection. CVPR 2020.
245 BirdNet+ (legacy)
This method makes use of Velodyne laser scans.
 code 51.85 % 70.14 % 50.03 % 0.1 s Titan Xp (PyTorch)
A. Barrera, C. Guindel, J. Beltrán and F. García: BirdNet+: End-to-End 3D Object Detection in LiDAR Bird’s Eye View. 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC) 2020.
246 ppt 50.41 % 54.19 % 45.14 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
247 SOD 48.69 % 70.90 % 40.12 % 0.1 s 1 core @ 2.5 Ghz (Python)
248 Complexer-YOLO
This method makes use of Velodyne laser scans.
 47.34 % 55.93 % 42.60 % 0.06 s GPU @ 3.5 Ghz (C/C++)
M. Simon, K. Amende, A. Kraus, J. Honer, T. Samann, H. Kaulbersch, S. Milz and H. Michael Gross: Complexer-YOLO: Real-Time 3D Object Detection and Tracking on Semantic Point Clouds. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops 2019.
249 EGFN
This method uses stereo information.
 46.39 % 65.80 % 38.42 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
250 Disp R-CNN (velo)
This method uses stereo information.
 code 45.78 % 68.21 % 37.73 % 0.387 s GPU @ 2.5 Ghz (Python + C/C++)
J. Sun, L. Chen, Y. Xie, S. Zhang, Q. Jiang, X. Zhou and H. Bao: Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation. CVPR 2020.
251 CDN-PL++
This method uses stereo information.
 44.86 % 64.31 % 38.11 % 0.4 s GPU @ 2.5 Ghz (C/C++)
D. Garg, Y. Wang, B. Hariharan, M. Campbell, K. Weinberger and W. Chao: Wasserstein Distances for Stereo Disparity Estimation. Advances in Neural Information Processing Systems 2020.
252 Disp R-CNN
This method uses stereo information.
 code 43.27 % 67.02 % 36.43 % 0.387 s GPU @ 2.5 Ghz (Python + C/C++)
J. Sun, L. Chen, Y. Xie, S. Zhang, Q. Jiang, X. Zhou and H. Bao: Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation. CVPR 2020.
253 R-AGNO-Net 42.79 % 49.49 % 39.31 % 0.15 s 1 core @ 2.5 Ghz (C/C++)
254 Pseudo-LiDAR++
This method uses stereo information.
 code 42.43 % 61.11 % 36.99 % 0.4 s GPU @ 2.5 Ghz (Python)
Y. You, Y. Wang, W. Chao, D. Garg, G. Pleiss, B. Hariharan, M. Campbell and K. Weinberger: Pseudo-LiDAR++: Accurate Depth for 3D Object Detection in Autonomous Driving. International Conference on Learning Representations 2020.
255 OSE+ 41.60 % 62.67 % 35.82 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
256 YOLOStereo3D
This method uses stereo information.
 code 41.25 % 65.68 % 30.42 % 0.1 s GPU 1080Ti
Y. Liu, L. Wang and M. Liu: YOLOStereo3D: A Step Back to 2D for Efficient Stereo 3D Detection. 2021 International Conference on Robotics and Automation (ICRA) 2021.
257 BEVC 40.72 % 50.05 % 36.42 % 35ms GPU @ 1.5 Ghz (Python)
258 RT3D-GMP
This method uses stereo information.
 38.76 % 45.79 % 30.00 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
H. Königshof and C. Stiller: Learning-Based Shape Estimation with Grid Map Patches for Realtime 3D Object Detection for Automated Driving. 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC) 2020.
259 ZoomNet
This method uses stereo information.
 code 38.64 % 55.98 % 30.97 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
L. Z. Xu: ZoomNet: Part-Aware Adaptive Zooming Neural Network for 3D Object Detection. Proceedings of the AAAI Conference on Artificial Intelligence 2020.
260 OC Stereo
This method uses stereo information.
 code 37.60 % 55.15 % 30.25 % 0.35 s 1 core @ 2.5 Ghz (Python + C/C++)
A. Pon, J. Ku, C. Li and S. Waslander: Object-Centric Stereo Matching for 3D Object Detection. ICRA 2020.
261 Pseudo-Lidar
This method uses stereo information.
 code 34.05 % 54.53 % 28.25 % 0.4 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Wang, W. Chao, D. Garg, B. Hariharan, M. Campbell and K. Weinberger: Pseudo-LiDAR From Visual Depth Estimation: Bridging the Gap in 3D Object Detection for Autonomous Driving. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
262 Stereo CenterNet
This method uses stereo information.
 31.30 % 49.94 % 25.62 % 0.04 s GPU @ 2.5 Ghz (Python)
Y. Shi, Y. Guo, Z. Mi and X. Li: Stereo CenterNet-based 3D object detection for autonomous driving. Neurocomputing 2022.
263 Stereo R-CNN
This method uses stereo information.
 code 30.23 % 47.58 % 23.72 % 0.3 s GPU @ 2.5 Ghz (Python)
P. Li, X. Chen and S. Shen: Stereo R-CNN based 3D Object Detection for Autonomous Driving. CVPR 2019.
264 BirdNet
This method makes use of Velodyne laser scans.
 27.26 % 40.99 % 25.32 % 0.11 s Titan Xp (Caffe)
J. Beltrán, C. Guindel, F. Moreno, D. Cruzado, F. García and A. Escalera: BirdNet: A 3D Object Detection Framework from LiDAR Information. 2018 21st International Conference on Intelligent Transportation Systems (ITSC) 2018.
265 GCDR 23.92 % 34.89 % 19.59 % 0.28 s 1 core @ 2.5 Ghz (Python)
266 RT3DStereo
This method uses stereo information.
 23.28 % 29.90 % 18.96 % 0.08 s GPU @ 2.5 Ghz (C/C++)
H. Königshof, N. Salscheider and C. Stiller: Realtime 3D Object Detection for Automated Driving Using Stereo Vision and Semantic Information. Proc. IEEE Intl. Conf. Intelligent Transportation Systems 2019.
267 Digging_M3D 21.24 % 29.15 % 19.18 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
268 RT3D
This method makes use of Velodyne laser scans.
 19.14 % 23.74 % 18.86 % 0.09 s GPU @ 1.8Ghz
Y. Zeng, Y. Hu, S. Liu, J. Ye, Y. Han, X. Li and N. Sun: RT3D: Real-Time 3-D Vehicle Detection in LiDAR Point Cloud for Autonomous Driving. IEEE Robotics and Automation Letters 2018.
269 Mix-Teaching-M3D 18.54 % 26.89 % 15.79 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
270 StereoFENet
This method uses stereo information.
 18.41 % 29.14 % 14.20 % 0.15 s 1 core @ 3.5 Ghz (Python)
W. Bao, B. Xu and Z. Chen: MonoFENet: Monocular 3D Object Detection with Feature Enhancement Networks. IEEE Transactions on Image Processing 2019.
271 CMKD 18.00 % 28.10 % 15.60 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
272 SCSTSV-MonoFlex 17.91 % 27.38 % 15.58 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
273 LPCG-Monoflex 17.80 % 25.56 % 15.38 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
274 PS-fld 17.74 % 23.74 % 15.14 % 0.25 s 1 core @ 2.5 Ghz (C/C++)
275 MonoDDE 17.14 % 24.93 % 15.10 % 0.04 s 1 core @ 2.5 Ghz (Python)
276 Mobile Stereo R-CNN
This method uses stereo information.
 17.04 % 26.97 % 13.26 % 1.8 s NVIDIA Jetson TX2
M. Hussein, M. Khalil and B. Abdullah: 3D Object Detection using Mobile Stereo R- CNN on Nvidia Jetson TX2. International Conference on Advanced Engineering, Technology and Applications (ICAETA) 2021.
277 CMKD 16.99 % 27.20 % 15.08 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
278 DD3D code 16.87 % 23.19 % 14.36 % n/a s 1 core @ 2.5 Ghz (C/C++)
D. Park, R. Ambrus, V. Guizilini, J. Li and A. Gaidon: Is Pseudo-Lidar needed for Monocular 3D Object detection?. IEEE/CVF International Conference on Computer Vision (ICCV) .
279 MonoCon code 16.46 % 22.50 % 13.95 % 0.02 s GPU @ 2.5 Ghz (Python)
T. Xianpeng Liu: Learning Auxiliary Monocular Contexts Helps Monocular 3D Object Detection. AAAI 2022.
280 Lite-FPN-GUPNet 16.20 % 23.58 % 13.56 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
281 gupnet_se 16.10 % 23.62 % 13.41 % 0.03s 1 core @ 2.5 Ghz (C/C++)
282 MonoDistill 16.03 % 22.97 % 13.60 % 0.04 s 1 core @ 2.5 Ghz (Python)
283 MDNet 16.01 % 24.59 % 13.49 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
284 mono3d code 15.73 % 23.96 % 13.35 % TBD TBD
285 MonoDTR 15.39 % 21.99 % 12.73 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
286 mono3d 15.26 % 23.41 % 12.80 % 0.03 s GPU @ 2.5 Ghz (Python)
287 HBD 15.17 % 21.71 % 13.06 % 0.12 s 1 core @ 2.5 Ghz (C/C++)
288 ZongmuMono3d code 15.08 % 23.79 % 13.25 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
289 GUPNet code 15.02 % 22.26 % 13.12 % NA s 1 core @ 2.5 Ghz (Python + C/C++)
Y. Lu, X. Ma, L. Yang, T. Zhang, Y. Liu, Q. Chu, J. Yan and W. Ouyang: Geometry Uncertainty Projection Network for Monocular 3D Object Detection. arXiv preprint arXiv:2107.13774 2021.
290 vadin-TBD 14.94 % 21.75 % 13.07 % 0.04 s 1 core @ 2.5 Ghz (Python)
291 LPCG-M3D 14.82 % 22.73 % 12.88 % 0.11 s 1 core @ 2.5 Ghz (C/C++)
292 M3DSSD++ code 14.75 % 23.61 % 11.80 % 0.16s 1 core @ 2.5 Ghz (C/C++)
293 MonoFlex 14.73 % 22.29 % 12.77 % 0.03 s 1 core @ 2.5 Ghz (Python)
294 SGM3D 14.65 % 22.46 % 12.97 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
Z. Zhou, L. Du, X. Ye, Z. Zou, X. Tan, E. Ding, L. Zhang, X. Xue and J. Feng: SGM3D: Stereo Guided Monocular 3D Object Detection. 2021.
295 Anonymous code 14.56 % 20.65 % 11.92 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
296 SAIC_ADC_Mono3D code 14.54 % 18.98 % 12.86 % 50 s GPU @ 2.5 Ghz (Python)
297 EW code 14.50 % 23.37 % 11.88 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
298 CA3D 14.49 % 20.89 % 12.19 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
299 MonoEdge 14.47 % 21.08 % 12.73 % 0.05 s GPU @ 2.5 Ghz (Python)
300 MonoGround 14.36 % 21.37 % 12.62 % 0.03 s 1 core @ 2.5 Ghz (Python)
301 ANM 14.33 % 20.84 % 11.61 % 0.02 s 1 core @ 2.5 Ghz (C/C++)
302 DLE code 14.33 % 24.23 % 10.30 % 0.06 s NVIDIA Tesla V100
C. Liu, S. Gu, L. Gool and R. Timofte: Deep Line Encoding for Monocular 3D Object Detection and Depth Prediction. Proceedings of the British Machine Vision Conference (BMVC) 2021.
303 ITS-MDPL 14.28 % 24.67 % 12.13 % 0.16 s GPU @ 2.5 Ghz (Python)
304 SwinMono3D 14.24 % 22.61 % 10.11 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
305 AutoShape code 14.17 % 22.47 % 11.36 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
Z. Liu, D. Zhou, F. Lu, J. Fang and L. Zhang: AutoShape: Real-Time Shape-Aware Monocular 3D Object Detection. Proceedings of the IEEE/CVF International Conference on Computer Vision 2021.
306 MonoEdge-Rotate 14.13 % 21.60 % 12.27 % 0.05 s GPU @ 2.5 Ghz (Python)
307 EM code 14.00 % 22.93 % 11.26 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
308 MAOLoss code 14.00 % 20.05 % 11.81 % 0.05 s 1 core @ 2.5 Ghz (Python)
309 E2E-DA 13.97 % 19.73 % 11.82 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
310 GAC3D++ 13.90 % 19.53 % 11.77 % 0.25 s 1 core @ 2.5 Ghz (C/C++)
311 MonoFlex 13.89 % 19.94 % 12.07 % 0.03 s GPU @ 2.5 Ghz (Python)
Y. Zhang, J. Lu and J. Zhou: Objects are Different: Flexible Monocular 3D Object Detection. CVPR 2021.
312 MonoEF code 13.87 % 21.29 % 11.71 % 0.03 s 1 core @ 2.5 Ghz (Python)
Y. Zhou, Y. He, H. Zhu, C. Wang, H. Li and Q. Jiang: Monocular 3D Object Detection: An Extrinsic Parameter Free Approach. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2021.
313 MonoGeo 13.81 % 18.85 % 11.52 % 0.05 s 1 core @ 2.5 Ghz (Python)
314 K3D 13.80 % 20.04 % 11.67 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
315 none 13.79 % 18.84 % 11.52 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
316 DFR-Net 13.63 % 19.40 % 10.35 % 0.18 s 1080 Ti (Pytorch)
Z. Zou, X. Ye, L. Du, X. Cheng, X. Tan, L. Zhang, J. Feng, X. Xue and E. Ding: The devil is in the task: Exploiting reciprocal appearance-localization features for monocular 3d object detection . ICCV 2021.
317 MonoLCD 13.52 % 18.08 % 11.58 % 0.04 s 1 core @ 2.5 Ghz (Python)
318 CaDDN code 13.41 % 19.17 % 11.46 % 0.63 s GPU @ 2.5 Ghz (Python)
C. Reading, A. Harakeh, J. Chae and S. Waslander: Categorical Depth Distribution Network for Monocular 3D Object Detection. CVPR 2021.
319 MDSNet 13.40 % 22.80 % 10.27 % 0.07 s 1 core @ 2.5 Ghz (Python)
320 PCT code 13.37 % 21.00 % 11.31 % 0.045 s 1 core @ 2.5 Ghz (C/C++)
L. Wang, L. Zhang, Y. Zhu, Z. Zhang, T. He, M. Li and X. Xue: Progressive Coordinate Transforms for Monocular 3D Object Detection. NeurIPS 2021.
321 KAIST-VDCLab 13.33 % 19.06 % 11.90 % 0.04 s 1 core @ 2.5 Ghz (Python)
322 Ground-Aware code 13.25 % 21.65 % 9.91 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
Y. Liu, Y. Yuan and M. Liu: Ground-aware Monocular 3D Object Detection for Autonomous Driving. IEEE Robotics and Automation Letters 2021.
323 MonoHMOO 13.12 % 20.28 % 9.56 % 0.2 s 1 core @ 2.5 Ghz (Python + C/C++)
324 Aug3D-RPN 12.99 % 17.82 % 9.78 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
C. He, J. Huang, X. Hua and L. Zhang: Aug3D-RPN: Improving Monocular 3D Object Detection by Synthetic Images with Virtual Depth. 2021.
325 vadin-TBD2 code 12.99 % 20.10 % 10.50 % 0.20 s 1 core @ 2.5 Ghz (Python)
326 RelationNet3D_dla34 code 12.88 % 17.67 % 11.01 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
327 DDMP-3D 12.78 % 19.71 % 9.80 % 0.18 s 1 core @ 2.5 Ghz (Python)
L. Wang, L. Du, X. Ye, Y. Fu, G. Guo, X. Xue, J. Feng and L. Zhang: Depth-conditioned Dynamic Message Propagation for Monocular 3D Object Detection. CVPR 2020.
328 RetinaMono 12.73 % 19.41 % 10.45 % 0.02 s 1 core @ 2.5 Ghz (Python)
329 Kinematic3D code 12.72 % 19.07 % 9.17 % 0.12 s 1 core @ 1.5 Ghz (C/C++)
G. Brazil, G. Pons-Moll, X. Liu and B. Schiele: Kinematic 3D Object Detection in Monocular Video. ECCV 2020 .
330 MonoRCNN code 12.65 % 18.36 % 10.03 % 0.07 s GPU @ 2.5 Ghz (Python)
X. Shi, Q. Ye, X. Chen, C. Chen, Z. Chen and T. Kim: Geometry-based Distance Decomposition for Monocular 3D Object Detection. ICCV 2021.
331 RelationNet3D 12.60 % 17.57 % 10.95 % 0.04 s GPU @ 2.5 Ghz (Python)
332 TBD 12.53 % 22.40 % 10.64 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
333 AutoShape 12.42 % 20.35 % 9.70 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
334 MP-Mono 12.37 % 17.89 % 9.58 % 0.16 s GPU @ 2.5 Ghz (Python)
335 GrooMeD-NMS code 12.32 % 18.10 % 9.65 % 0.12 s 1 core @ 2.5 Ghz (Python)
A. Kumar, G. Brazil and X. Liu: GrooMeD-NMS: Grouped Mathematically Differentiable NMS for Monocular 3D Object Detection. CVPR 2021.
336 MonoRUn code 12.30 % 19.65 % 10.58 % 0.07 s GPU @ 2.5 Ghz (Python + C/C++)
H. Chen, Y. Huang, W. Tian, Z. Gao and L. Xiong: MonoRUn: Monocular 3D Object Detection by Reconstruction and Uncertainty Propagation. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2021.
337 monodle code 12.26 % 17.23 % 10.29 % 0.04 s GPU @ 2.5 Ghz (Python)
X. Ma, Y. Zhang, D. Xu, D. Zhou, S. Yi, H. Li and W. Ouyang: Delving into Localization Errors for Monocular 3D Object Detection. CVPR 2021 .
338 LT-M3OD 12.26 % 18.15 % 10.05 % 0.03 s 1 core @ 2.5 Ghz (Python)
339 PPTrans 12.06 % 19.79 % 10.48 % 0.2 s GPU @ 2.5 Ghz (Python)
340 YoloMono3D code 12.06 % 18.28 % 8.42 % 0.05 s GPU @ 2.5 Ghz (Python)
Y. Liu, L. Wang and L. Ming: YOLOStereo3D: A Step Back to 2D for Efficient Stereo 3D Detection. 2021 International Conference on Robotics and Automation (ICRA) 2021.
341 IAFA 12.01 % 17.81 % 10.61 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
D. Zhou, X. Song, Y. Dai, J. Yin, F. Lu, M. Liao, J. Fang and L. Zhang: IAFA: Instance-Aware Feature Aggregation for 3D Object Detection from a Single Image. Proceedings of the Asian Conference on Computer Vision 2020.
342 GAC3D 12.00 % 17.75 % 9.15 % 0.25 s 1 core @ 2.5 Ghz (Python)
M. Bui, D. Ngo, H. Pham and D. Nguyen: GAC3D: improving monocular 3D object detection with ground-guide model and adaptive convolution. 2021.
343 PGD-FCOS3D code 11.76 % 19.05 % 9.39 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
T. Wang, X. Zhu, J. Pang and D. Lin: Probabilistic and Geometric Depth: Detecting Objects in Perspective. Conference on Robot Learning (CoRL) 2021.
344 D4LCN code 11.72 % 16.65 % 9.51 % 0.2 s GPU @ 2.5 Ghz (Python + C/C++)
M. Ding, Y. Huo, H. Yi, Z. Wang, J. Shi, Z. Lu and P. Luo: Learning Depth-Guided Convolutions for Monocular 3D Object Detection. CVPR 2020.
345 KM3D code 11.45 % 16.73 % 9.92 % 0.03 s 1 core @ 2.5 Ghz (Python)
P. Li: Monocular 3D Detection with Geometric Constraints Embedding and Semi-supervised Training. 2020.
346 RefinedMPL 11.14 % 18.09 % 8.94 % 0.15 s GPU @ 2.5 Ghz (Python + C/C++)
J. Vianney, S. Aich and B. Liu: RefinedMPL: Refined Monocular PseudoLiDAR for 3D Object Detection in Autonomous Driving. arXiv preprint arXiv:1911.09712 2019.
347 PatchNet code 11.12 % 15.68 % 10.17 % 0.4 s 1 core @ 2.5 Ghz (C/C++)
X. Ma, S. Liu, Z. Xia, H. Zhang, X. Zeng and W. Ouyang: Rethinking Pseudo-LiDAR Representation. Proceedings of the European Conference on Computer Vision (ECCV) 2020.
348 ImVoxelNet code 10.97 % 17.15 % 9.15 % 0.2 s GPU @ 2.5 Ghz (Python)
D. Rukhovich, A. Vorontsova and A. Konushin: ImVoxelNet: Image to Voxels Projection for Monocular and Multi-View General-Purpose 3D Object Detection. arXiv preprint arXiv:2106.01178 2021.
349 COF3D 10.91 % 17.86 % 8.20 % 200 s 1 core @ 2.5 Ghz (Python + C/C++)
350 MM 10.74 % 15.80 % 8.64 % 1 s 1 core @ 2.5 Ghz (C/C++)
351 AM3D 10.74 % 16.50 % 9.52 % 0.4 s GPU @ 2.5 Ghz (Python + C/C++)
X. Ma, Z. Wang, H. Li, P. Zhang, W. Ouyang and X. Fan: Accurate Monocular Object Detection via Color- Embedded 3D Reconstruction for Autonomous Driving. Proceedings of the IEEE international Conference on Computer Vision (ICCV) 2019.
352 Lite-FPN 10.64 % 15.32 % 8.59 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
353 TBD 10.61 % 15.71 % 8.66 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
354 Keypoint-3D 10.42 % 15.97 % 7.91 % 14 s 1 core @ 2.5 Ghz (C/C++)
355 MK3D 10.37 % 15.39 % 8.82 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
356 RTM3D code 10.34 % 14.41 % 8.77 % 0.05 s GPU @ 1.0 Ghz (Python)
P. Li, H. Zhao, P. Liu and F. Cao: RTM3D: Real-time Monocular 3D Detection from Object Keypoints for Autonomous Driving. 2020.
357 E2E-DA-Lite (Res18) 10.32 % 15.56 % 8.89 % 0.01 s GPU @ 2.5 Ghz (Python)
358 MonoPair 9.99 % 13.04 % 8.65 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Chen, L. Tai, K. Sun and M. Li: MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationships. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020.
359 RelationNet3D_res18 code 9.93 % 14.27 % 8.43 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
360 Neighbor-Vote 9.90 % 15.57 % 8.89 % 0.1 s GPU @ 2.5 Ghz (Python)
X. Chu, J. Deng, Y. Li, Z. Yuan, Y. Zhang, J. Ji and Y. Zhang: Neighbor-Vote: Improving Monocular 3D Object Detection through Neighbor Distance Voting. ACM MM 2021.
361 SMOKE code 9.76 % 14.03 % 7.84 % 0.03 s GPU @ 2.5 Ghz (Python)
Z. Liu, Z. Wu and R. Tóth: SMOKE: Single-Stage Monocular 3D Object Detection via Keypoint Estimation. 2020.
362 M3D-RPN code 9.71 % 14.76 % 7.42 % 0.16 s GPU @ 1.5 Ghz (Python)
G. Brazil and X. Liu: M3D-RPN: Monocular 3D Region Proposal Network for Object Detection . ICCV 2019 .
363 QD-3DT
This is an online method (no batch processing).
 code 9.33 % 12.81 % 7.86 % 0.03 s GPU @ 2.5 Ghz (Python)
H. Hu, Y. Yang, T. Fischer, F. Yu, T. Darrell and M. Sun: Monocular Quasi-Dense 3D Object Tracking. ArXiv:2103.07351 2021.
364 ICCV 9.31 % 13.37 % 8.29 % 0.04 s GPU @ 2.5 Ghz (Python)
365 TopNet-HighRes
This method makes use of Velodyne laser scans.
 9.28 % 12.67 % 7.95 % 101ms NVIDIA GeForce 1080 Ti (tensorflow-gpu)
S. Wirges, T. Fischer, C. Stiller and J. Frias: Object Detection and Classification in Occupancy Grid Maps Using Deep Convolutional Networks. 2018 21st International Conference on Intelligent Transportation Systems (ITSC) 2018.
366 MonoCInIS 7.94 % 15.82 % 6.68 % 0,13 s GPU @ 2.5 Ghz (C/C++)
J. Heylen, M. De Wolf, B. Dawagne, M. Proesmans, L. Van Gool, W. Abbeloos, H. Abdelkawy and D. Reino: MonoCInIS: Camera Independent Monocular 3D Object Detection using Instance Segmentation. Proceedings of the IEEE/CVF International Conference on Computer Vision 2021.
367 Geo3D 7.70 % 11.52 % 6.80 % 0.04 s GPU @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
368 SS3D 7.68 % 10.78 % 6.51 % 48 ms Tesla V100 (Python)
E. Jörgensen, C. Zach and F. Kahl: Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss. CoRR 2019.
369 MonoCInIS 7.66 % 15.21 % 6.24 % 0,14 s GPU @ 2.5 Ghz (Python)
J. Heylen, M. De Wolf, B. Dawagne, M. Proesmans, L. Van Gool, W. Abbeloos, H. Abdelkawy and D. Reino: MonoCInIS: Camera Independent Monocular 3D Object Detection using Instance Segmentation. Proceedings of the IEEE/CVF International Conference on Computer Vision 2021.
370 Mono3D_PLiDAR code 7.50 % 10.76 % 6.10 % 0.1 s NVIDIA GeForce 1080 (pytorch)
X. Weng and K. Kitani: Monocular 3D Object Detection with Pseudo-LiDAR Point Cloud. arXiv:1903.09847 2019.
371 MonoPSR code 7.25 % 10.76 % 5.85 % 0.2 s GPU @ 3.5 Ghz (Python)
J. Ku*, A. Pon* and S. Waslander: Monocular 3D Object Detection Leveraging Accurate Proposals and Shape Reconstruction. CVPR 2019.
372 Decoupled-3D 7.02 % 11.08 % 5.63 % 0.08 s GPU @ 2.5 Ghz (C/C++)
Y. Cai, B. Li, Z. Jiao, H. Li, X. Zeng and X. Wang: Monocular 3D Object Detection with Decoupled Structured Polygon Estimation and Height-Guided Depth Estimation. AAAI 2020.
373 VoxelJones code 6.35 % 7.39 % 5.80 % .18 s 1 core @ 2.5 Ghz (Python + C/C++)
M. Motro and J. Ghosh: Vehicular Multi-object Tracking with Persistent Detector Failures. arXiv preprint arXiv:1907.11306 2019.
374 MonoGRNet code 5.74 % 9.61 % 4.25 % 0.04s NVIDIA P40
Z. Qin, J. Wang and Y. Lu: MonoGRNet: A Geometric Reasoning Network for 3D Object Localization. The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19) 2019.
375 A3DODWTDA (image) code 5.27 % 6.88 % 4.45 % 0.8 s GPU @ 3.0 Ghz (Python)
F. Gustafsson and E. Linder-Norén: Automotive 3D Object Detection Without Target Domain Annotations. 2018.
376 MonoFENet 5.14 % 8.35 % 4.10 % 0.15 s 1 core @ 3.5 Ghz (Python)
W. Bao, B. Xu and Z. Chen: MonoFENet: Monocular 3D Object Detection with Feature Enhancement Networks. IEEE Transactions on Image Processing 2019.
377 TLNet (Stereo)
This method uses stereo information.
 code 4.37 % 7.64 % 3.74 % 0.1 s 1 core @ 2.5 Ghz (Python)
Z. Qin, J. Wang and Y. Lu: Triangulation Learning Network: from Monocular to Stereo 3D Object Detection. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
378 CSoR
This method makes use of Velodyne laser scans.
 4.06 % 5.61 % 3.17 % 3.5 s 4 cores @ >3.5 Ghz (Python + C/C++)
L. Plotkin: PyDriver: Entwicklung eines Frameworks für räumliche Detektion und Klassifikation von Objekten in Fahrzeugumgebung. 2015.
379 Shift R-CNN (mono) code 3.87 % 6.88 % 2.83 % 0.25 s GPU @ 1.5 Ghz (Python)
A. Naiden, V. Paunescu, G. Kim, B. Jeon and M. Leordeanu: Shift R-CNN: Deep Monocular 3D Object Detection With Closed-form Geometric Constraints. ICIP 2019.
380 MVRA + I-FRCNN+ 3.27 % 5.19 % 2.49 % 0.18 s GPU @ 2.5 Ghz (Python)
H. Choi, H. Kang and Y. Hyun: Multi-View Reprojection Architecture for Orientation Estimation. The IEEE International Conference on Computer Vision (ICCV) Workshops 2019.
381 SparVox3D 3.20 % 5.27 % 2.56 % 0.05 s GPU @ 2.0 Ghz (Python)
E. Balatkan and F. Kıraç: Improving Regression Performance on Monocular 3D Object Detection Using Bin-Mixing and Sparse Voxel Data. 2021 6th International Conference on Computer Science and Engineering (UBMK) 2021.
382 TopNet-UncEst
This method makes use of Velodyne laser scans.
 3.02 % 3.24 % 2.26 % 0.09 s NVIDIA GeForce 1080 Ti (tensorflow-gpu)
S. Wirges, M. Braun, M. Lauer and C. Stiller: Capturing Object Detection Uncertainty in Multi-Layer Grid Maps. 2019.
383 GS3D 2.90 % 4.47 % 2.47 % 2 s 1 core @ 2.5 Ghz (C/C++)
B. Li, W. Ouyang, L. Sheng, X. Zeng and X. Wang: GS3D: An Efficient 3D Object Detection Framework for Autonomous Driving. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
384 3D-GCK 2.52 % 3.27 % 2.11 % 24 ms Tesla V100
N. Gählert, J. Wan, N. Jourdan, J. Finkbeiner, U. Franke and J. Denzler: Single-Shot 3D Detection of Vehicles from Monocular RGB Images via Geometrically Constrained Keypoints in Real-Time. 2020 IEEE Intelligent Vehicles Symposium (IV) 2020.
385 WeakM3D 2.26 % 5.03 % 1.63 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
386 ROI-10D 2.02 % 4.32 % 1.46 % 0.2 s GPU @ 3.5 Ghz (Python)
F. Manhardt, W. Kehl and A. Gaidon: ROI-10D: Monocular Lifting of 2D Detection to 6D Pose and Metric Shape. Computer Vision and Pattern Recognition (CVPR) 2019.
387 CDTrack3D code 1.65 % 3.01 % 1.40 % 0.0106 s NVIDIA RTX 3090 GPU, i9 10850k CPU
388 FQNet 1.51 % 2.77 % 1.01 % 0.5 s 1 core @ 2.5 Ghz (Python)
L. Liu, J. Lu, C. Xu, Q. Tian and J. Zhou: Deep Fitting Degree Scoring Network for Monocular 3D Object Detection. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2019.
389 3D-SSMFCNN code 1.41 % 1.88 % 1.11 % 0.1 s GPU @ 1.5 Ghz (C/C++)
L. Novak: Vehicle Detection and Pose Estimation for Autonomous Driving. 2017.
390 test code 0.03 % 0.01 % 0.03 % 50 s 1 core @ 2.5 Ghz (Python)
391 mBoW
This method makes use of Velodyne laser scans.
 0.00 % 0.00 % 0.00 % 10 s 1 core @ 2.5 Ghz (C/C++)
J. Behley, V. Steinhage and A. Cremers: Laser-based Segment Classification Using a Mixture of Bag-of-Words. Proc. of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2013.
Table as LaTeX | Only published Methods

Pedestrian


Method Setting Code Moderate Easy Hard Runtime Environment
1 VPFNet code 48.36 % 54.65 % 44.98 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
C. Wang, H. Chen and L. Fu: VPFNet: Voxel-Pixel Fusion Network for Multi-class 3D Object Detection. 2021.
2 CAD 47.91 % 55.98 % 44.63 % 0.1 s GPU @ 2.5 Ghz (Python + C/C++)
3 PV-RCNN++ 47.19 % 54.29 % 43.49 % 0.06 s 1 core @ 2.5 Ghz (C/C++)
4 CasA 47.09 % 54.04 % 44.56 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
5 EQ-PVRCNN 47.02 % 55.84 % 42.94 % 0.2 s GPU @ 2.5 Ghz (Python + C/C++)
6 ADLAB 46.18 % 53.59 % 43.28 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
7 PiFeNet 45.89 % 54.84 % 42.02 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
D. Le, H. Shi, H. Rezatofighi and J. Cai: PiFeNet: Pillar-Feature Network for Real-Time 3D Pedestrian Detection from Point Cloud. 2021.
8 ISE-RCNN 45.66 % 51.44 % 42.43 % 0.09 s 1 core @ 2.5 Ghz (Python + C/C++)
9 CAT-Det 45.44 % 54.26 % 41.94 % 0.3 s GPU @ 2.5 Ghz (Python + C/C++)
10 HotSpotNet 45.37 % 53.10 % 41.47 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
Q. Chen, L. Sun, Z. Wang, K. Jia and A. Yuille: object as hotspots. Proceedings of the European Conference on Computer Vision (ECCV) 2020.
11 H^23D R-CNN 45.26 % 52.75 % 41.56 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
12 PE-RCVN 45.01 % 50.29 % 41.85 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
13 SAA-PV-RCNN 45.00 % 52.55 % 41.82 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
14 VPN 44.56 % 54.13 % 41.73 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
15 EPNet++ 44.38 % 52.79 % 41.29 % 0.1 s GPU @ 2.5 Ghz (Python)
16 TANet code 44.34 % 53.72 % 40.49 % 0.035s GPU @ 2.5 Ghz (Python + C/C++)
Z. Liu, X. Zhao, T. Huang, R. Hu, Y. Zhou and X. Bai: TANet: Robust 3D Object Detection from Point Clouds with Triple Attention. AAAI 2020.
17 TBD 44.32 % 49.37 % 41.24 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
18 3DSSD code 44.27 % 54.64 % 40.23 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Yang, Y. Sun, S. Liu and J. Jia: 3DSSD: Point-based 3D Single Stage Object Detector. CVPR 2020.
19 AutoAlign 44.08 % 53.99 % 40.82 % 0.1 s 1 core @ 2.5 Ghz (Python)
20 ISE-RCNN-PV 43.78 % 50.03 % 40.50 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
21 Point-GNN
This method makes use of Velodyne laser scans.
 code 43.77 % 51.92 % 40.14 % 0.6 s GPU @ 2.5 Ghz (Python)
W. Shi and R. Rajkumar: Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud. CVPR 2020.
22 VCT 43.65 % 50.27 % 41.43 % 0.2 s 1 core @ 2.5 Ghz (Python)
23 USVLab BSAODet (MM) 43.63 % 51.71 % 41.09 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
24 EA-M-RCNN(BorderAtt) 43.44 % 51.81 % 39.85 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
25 F-ConvNet
This method makes use of Velodyne laser scans.
 code 43.38 % 52.16 % 38.80 % 0.47 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Wang and K. Jia: Frustum ConvNet: Sliding Frustums to Aggregate Local Point-Wise Features for Amodal 3D Object Detection. IROS 2019.
26 MMLab-PartA^2
This method makes use of Velodyne laser scans.
 code 43.35 % 53.10 % 40.06 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
S. Shi, Z. Wang, J. Shi, X. Wang and H. Li: From Points to Parts: 3D Object Detection from Point Cloud with Part-aware and Part-aggregation Network. IEEE Transactions on Pattern Analysis and Machine Intelligence 2020.
27 FS-Net
This method makes use of Velodyne laser scans.
 43.31 % 49.82 % 40.89 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
28 MMLab PV-RCNN
This method makes use of Velodyne laser scans.
 code 43.29 % 52.17 % 40.29 % 0.08 s 1 core @ 2.5 Ghz (Python + C/C++)
S. Shi, C. Guo, L. Jiang, Z. Wang, J. Shi, X. Wang and H. Li: PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detection. CVPR 2020.
29 FromVoxelToPoint code 43.28 % 51.80 % 40.71 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
J. Li, H. Dai, L. Shao and Y. Ding: From Voxel to Point: IoU-guided 3D Object Detection for Point Cloud with Voxel-to- Point Decoder. MM '21: The 29th ACM International Conference on Multimedia (ACM MM) 2021.
30 VMVS
This method makes use of Velodyne laser scans.
 43.27 % 53.44 % 39.51 % 0.25 s GPU @ 2.5 Ghz (Python)
J. Ku, A. Pon, S. Walsh and S. Waslander: Improving 3D object detection for pedestrians with virtual multi-view synthesis orientation estimation. IROS 2019.
31 P2V-RCNN 43.19 % 50.91 % 40.81 % 0.1 s 2 cores @ 2.5 Ghz (Python)
J. Li, S. Luo, Z. Zhu, H. Dai, A. Krylov, Y. Ding and L. Shao: P2V-RCNN: Point to Voxel Feature Learning for 3D Object Detection from Point Clouds. IEEE Access 2021.
32 MGAF-3DSSD code 43.09 % 50.65 % 39.65 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
J. Li, H. Dai, L. Shao and Y. Ding: Anchor-free 3D Single Stage Detector with Mask-Guided Attention for Point Cloud. MM '21: The 29th ACM International Conference on Multimedia (ACM MM) 2021.
33 SGNet 43.00 % 49.68 % 40.45 % 0.09 s GPU @ 2.5 Ghz (Python)
34 Frustum-PointPillars code 42.89 % 51.22 % 39.28 % 0.06 s 4 cores @ 3.0 Ghz (Python)
A. Paigwar, D. Sierra-Gonzalez, \. Erkent and C. Laugier: Frustum-PointPillars: A Multi-Stage Approach for 3D Object Detection using RGB Camera and LiDAR. International Conference on Computer Vision, ICCV, Workshop on Autonomous Vehicle Vision 2021.
35 Fast-CLOCs 42.72 % 52.10 % 39.08 % 0.1 s GPU @ 2.5 Ghz (Python)
S. Pang, D. Morris and H. Radha: Fast-CLOCs: Fast Camera-LiDAR Object Candidates Fusion for 3D Object Detection. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2022.
36 TBD 42.65 % 50.88 % 39.78 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
37 USVLab BSAODet (SM) 42.62 % 49.52 % 39.12 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
38 STD code 42.47 % 53.29 % 38.35 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Yang, Y. Sun, S. Liu, X. Shen and J. Jia: STD: Sparse-to-Dense 3D Object Detector for Point Cloud. ICCV 2019.
39 AVOD-FPN
This method makes use of Velodyne laser scans.
 code 42.27 % 50.46 % 39.04 % 0.1 s Titan X (Pascal)
J. Ku, M. Mozifian, J. Lee, A. Harakeh and S. Waslander: Joint 3D Proposal Generation and Object Detection from View Aggregation. IROS 2018.
40 SemanticVoxels 42.19 % 50.90 % 39.52 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
J. Fei, W. Chen, P. Heidenreich, S. Wirges and C. Stiller: SemanticVoxels: Sequential Fusion for 3D Pedestrian Detection using LiDAR Point Cloud and Semantic Segmentation. MFI 2020.
41 TBD 42.19 % 49.89 % 39.34 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
42 TBD 42.19 % 49.89 % 39.34 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
43 F-PointNet
This method makes use of Velodyne laser scans.
 code 42.15 % 50.53 % 38.08 % 0.17 s GPU @ 3.0 Ghz (Python)
C. Qi, W. Liu, C. Wu, H. Su and L. Guibas: Frustum PointNets for 3D Object Detection from RGB-D Data. arXiv preprint arXiv:1711.08488 2017.
44 TBD
This method makes use of Velodyne laser scans.
 42.05 % 48.66 % 38.94 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
45 PointPillars
This method makes use of Velodyne laser scans.
 code 41.92 % 51.45 % 38.89 % 16 ms 1080ti GPU and Intel i7 CPU
A. Lang, S. Vora, H. Caesar, L. Zhou, J. Yang and O. Beijbom: PointPillars: Fast Encoders for Object Detection from Point Clouds. CVPR 2019.
46 TBD_IOU1 41.65 % 49.00 % 39.31 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
47 epBRM
This method makes use of Velodyne laser scans.
 code 41.52 % 49.17 % 39.08 % 0.10 s 1 core @ 2.5 Ghz (C/C++)
K. Shin: Improving a Quality of 3D Object Detection by Spatial Transformation Mechanism. arXiv preprint arXiv:1910.04853 2019.
48 TCDVF 41.47 % 49.44 % 38.41 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
49 TBD_IOU 41.45 % 48.25 % 39.17 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
50 DGT-Det3D 41.40 % 49.06 % 38.43 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
51 tbd 41.10 % 50.56 % 37.49 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
52 IA-SSD (single) 41.03 % 47.90 % 37.98 % 0.013 s 1 core @ 2.5 Ghz (C/C++)
53 Generalized-SIENet 40.97 % 47.01 % 38.88 % 0.08 s 1 core @ 2.5 Ghz (Python)
54 PointPainting
This method makes use of Velodyne laser scans.
 40.97 % 50.32 % 37.87 % 0.4 s GPU @ 2.5 Ghz (Python + C/C++)
S. Vora, A. Lang, B. Helou and O. Beijbom: PointPainting: Sequential Fusion for 3D Object Detection. CVPR 2020.
55 SCIR-Net
This method makes use of Velodyne laser scans.
 40.95 % 49.23 % 38.47 % 0.03 s GPU @ 2.5 Ghz (Python + C/C++)
56 DSA-PV-RCNN
This method makes use of Velodyne laser scans.
 code 40.89 % 46.97 % 38.80 % 0.08 s 1 core @ 2.5 Ghz (Python + C/C++)
P. Bhattacharyya, C. Huang and K. Czarnecki: SA-Det3D: Self-Attention Based Context-Aware 3D Object Detection. 2021.
57 SARFE 40.79 % 47.29 % 38.25 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
58 MSADet 40.58 % 49.54 % 38.19 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
59 SAA-SECOND 40.57 % 48.73 % 37.77 % 38m s 1 core @ 2.5 Ghz (C/C++)
60 PDV 40.56 % 47.80 % 38.46 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
61 WHUT-iou_ssd code 40.53 % 46.41 % 38.48 % 0.045s 1 core @ 2.5 Ghz (C/C++)
62 SA-voxel-centernet code 40.43 % 46.10 % 38.32 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
63 SVGA-Net 40.39 % 48.48 % 37.92 % 0.03s 1 core @ 2.5 Ghz (Python + C/C++)
Q. He, Z. Wang, H. Zeng, Y. Zeng and Y. Liu: SVGA-Net: Sparse Voxel-Graph Attention Network for 3D Object Detection from Point Clouds. AAAI 2022.
64 FusionDetv2-v3 40.38 % 46.86 % 37.41 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
65 FPCR-CNN 40.32 % 48.33 % 37.66 % 0.05 s 1 core @ 2.5 Ghz (Python)
66 P2V_PCV1 40.27 % 45.43 % 38.24 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
67 sa-voxel-centernet code 40.24 % 46.08 % 38.07 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
68 FPC-RCNN 40.13 % 46.41 % 37.84 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
69 TBD 40.07 % 46.11 % 37.87 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
70 TPCG 39.97 % 46.35 % 37.66 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
71 M3DeTR code 39.94 % 45.70 % 37.66 % n/a s GPU @ 1.0 Ghz (Python)
T. Guan, J. Wang, S. Lan, R. Chandra, Z. Wu, L. Davis and D. Manocha: M3DeTR: Multi-representation, Multi- scale, Mutual-relation 3D Object Detection with Transformers. 2021.
72 FusionDetv2-v5 39.91 % 47.50 % 37.39 % 0.05 s 1 core @ 2.5 Ghz (Java + C/C++)
73 DDet 39.87 % 45.82 % 38.00 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
74 MVOD 39.82 % 46.22 % 37.56 % 0.16 s 1 core @ 2.5 Ghz (C/C++)
75 Point Image Fusion 39.79 % 45.04 % 37.62 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
76 anonymous code 39.74 % 46.09 % 37.41 % 0.05s 1 core @ >3.5 Ghz (python)
77 FusionDetv2-v4 39.68 % 46.93 % 37.31 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
78 DSASNet 39.65 % 47.14 % 37.05 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
79 Fast VP-RCNN code 39.65 % 45.95 % 37.29 % 0.05 s 1 core @ 3.5 Ghz (C/C++)
80 VCRCNN 39.64 % 45.19 % 37.55 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
81 Dune-DCF-e09 39.43 % 47.29 % 36.74 % 1 s 1 core @ 2.5 Ghz (C/C++)
82 LazyTorch-CP-Infer-O 39.43 % 47.38 % 36.96 % 1 s 1 core @ 2.5 Ghz (C/C++)
83 SRDL 39.43 % 47.30 % 36.99 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
ERROR: Wrong syntax in BIBTEX file.
84 FusionDetv1 39.42 % 47.30 % 36.97 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
85 demo 39.38 % 47.69 % 36.06 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
86 MMLab-PointRCNN
This method makes use of Velodyne laser scans.
 code 39.37 % 47.98 % 36.01 % 0.1 s GPU @ 2.5 Ghz (Python + C/C++)
S. Shi, X. Wang and H. Li: Pointrcnn: 3d object proposal generation and detection from point cloud. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2019.
87 ST-RCNN
This method makes use of Velodyne laser scans.
 39.36 % 44.96 % 37.09 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
88 LazyTorch-CP-Small-P 39.33 % 47.27 % 36.89 % 1 s 1 core @ 2.5 Ghz (C/C++)
89 FusionDetv2-v2 39.31 % 44.98 % 37.22 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
90 ARPNET 39.31 % 48.32 % 35.93 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Ye, C. Zhang and X. Hao: ARPNET: attention region proposal network for 3D object detection. Science China Information Sciences 2019.
91 LazyTorch-CP 39.28 % 47.25 % 36.78 % 1 s 1 core @ 2.5 Ghz (C/C++)
92 Dune-DCF-e11 39.26 % 47.32 % 36.63 % 1 s 1 core @ 2.5 Ghz (C/C++)
93 City-CF-fixed 39.22 % 47.68 % 36.55 % 1 s 1 core @ 2.5 Ghz (C/C++)
94 IA-SSD (multi) 39.03 % 46.51 % 35.61 % 0.014 s 1 core @ 2.5 Ghz (C/C++)
95 NV-RCNN 38.75 % 47.05 % 36.52 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
96 SIF 38.74 % 46.23 % 36.06 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
P. An: SIF. Submitted to CVIU 2021.
97 CrazyTensor-CP 38.67 % 46.58 % 36.15 % 1 s 1 core @ 2.5 Ghz (Python)
98 SCNet
This method makes use of Velodyne laser scans.
 38.66 % 47.83 % 35.70 % 0.04 s GPU @ 3.0 Ghz (Python)
Z. Wang, H. Fu, L. Wang, L. Xiao and B. Dai: SCNet: Subdivision Coding Network for Object Detection Based on 3D Point Cloud. IEEE Access 2019.
99 Dune-DCF-e15 38.61 % 46.41 % 36.02 % 1 s 1 core @ 2.5 Ghz (C/C++)
100 Faraway-Frustum
This method makes use of Velodyne laser scans.
 code 38.58 % 46.33 % 35.71 % 0.1 s GPU @ 2.5 Ghz (Python)
H. Zhang, D. Yang, E. Yurtsever, K. Redmill and U. Ozguner: Faraway-Frustum: Dealing with Lidar Sparsity for 3D Object Detection using Fusion. arXiv preprint arXiv:2011.01404 2020.
101 FPV-SSD 38.45 % 45.83 % 36.03 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
102 TBD 38.27 % 46.35 % 36.39 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
103 City-CF 38.04 % 45.42 % 35.40 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
104 FPC3D_all
This method makes use of Velodyne laser scans.
 37.95 % 45.49 % 35.60 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
105 DVFENet 37.50 % 43.55 % 35.33 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
Y. He, G. Xia, Y. Luo, L. Su, Z. Zhang, W. Li and P. Wang: DVFENet: Dual-branch Voxel Feature Extraction Network for 3D Object Detection. Neurocomputing 2021.
106 MLOD
This method makes use of Velodyne laser scans.
 code 37.47 % 47.58 % 35.07 % 0.12 s GPU @ 1.5 Ghz (Python)
J. Deng and K. Czarnecki: MLOD: A multi-view 3D object detection based on robust feature fusion method. arXiv preprint arXiv:1909.04163 2019.
107 S-AT GCN 37.37 % 44.63 % 34.92 % 0.02 s GPU @ 2.0 Ghz (Python)
L. Wang, C. Wang, X. Zhang, T. Lan and J. Li: S-AT GCN: Spatial-Attention Graph Convolution Network based Feature Enhancement for 3D Object Detection. CoRR 2021.
108 YF 36.99 % 44.43 % 34.40 % 0.04 s GPU @ 2.5 Ghz (C/C++)
109 HS3D code 36.86 % 45.62 % 33.67 % 0.12 s 1 core @ 2.5 Ghz (Python + C/C++)
110 XView 36.79 % 42.44 % 34.96 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
L. Xie, G. Xu, D. Cai and X. He: X-view: Non-egocentric Multi-View 3D Object Detector. 2021.
111 FusionDetv2-baseline 36.66 % 41.34 % 34.60 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
112 TBD 36.53 % 44.11 % 34.30 % TBD GPU @ 2.5 Ghz (Python + C/C++)
113 PFF3D
This method makes use of Velodyne laser scans.
 code 36.07 % 43.93 % 32.86 % 0.05 s GPU @ 3.0 Ghz (Python + C/C++)
L. Wen and K. Jo: Fast and Accurate 3D Object Detection for Lidar-Camera-Based Autonomous Vehicles Using One Shared Voxel-Based Backbone. IEEE Access 2021.
114 NV2P-RCNN 35.98 % 43.18 % 33.88 % 0.1 s GPU @ 2.5 Ghz (Python)
115 CrazyTensor-CF 35.83 % 43.50 % 33.23 % 1 s 1 core @ 2.5 Ghz (C/C++)
116 ASCNet 35.76 % 42.00 % 33.69 % 0.09 s 1 core @ 2.5 Ghz (C/C++)
117 BirdNet+
This method makes use of Velodyne laser scans.
 code 35.06 % 41.55 % 32.93 % 0.11 s Titan Xp (PyTorch)
A. Barrera, J. Beltrán, C. Guindel, J. Iglesias and F. García: BirdNet+: Two-Stage 3D Object Detection in LiDAR through a Sparsity-Invariant Bird’s Eye View. IEEE Access 2021.
118 TBD_BD code 34.86 % 42.56 % 32.49 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
119 AB3DMOT
This method makes use of Velodyne laser scans.
This is an online method (no batch processing).
 code 34.59 % 42.27 % 31.37 % 0.0047s 1 core @ 2.5 Ghz (python)
X. Weng and K. Kitani: A Baseline for 3D Multi-Object Tracking. arXiv:1907.03961 2019.
120 DSGN++
This method uses stereo information.
 32.74 % 43.05 % 29.54 % 0.4 s NVIDIA Tesla V100
121 PP-PCdet code 32.04 % 39.23 % 29.79 % 0.01 s 1 core @ 2.5 Ghz (Python)
122 Contrastive PP code 31.64 % 38.47 % 29.30 % 0.01 s 1 core @ 2.5 Ghz (Python)
123 BirdNet+ (legacy)
This method makes use of Velodyne laser scans.
 code 31.46 % 37.99 % 29.46 % 0.1 s Titan Xp (PyTorch)
A. Barrera, C. Guindel, J. Beltrán and F. García: BirdNet+: End-to-End 3D Object Detection in LiDAR Bird’s Eye View. 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC) 2020.
124 SparsePool code 30.38 % 37.84 % 26.94 % 0.13 s 8 cores @ 2.5 Ghz (Python)
Z. Wang, W. Zhan and M. Tomizuka: Fusing bird view lidar point cloud and front view camera image for deep object detection. arXiv preprint arXiv:1711.06703 2017.
125 MMLAB LIGA-Stereo
This method uses stereo information.
 code 30.00 % 40.46 % 27.07 % 0.4 s 1 core @ 2.5 Ghz (Python + C/C++)
X. Guo, S. Shi, X. Wang and H. Li: LIGA-Stereo: Learning LiDAR Geometry Aware Representations for Stereo-based 3D Detector. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) 2021.
126 DisposalNet
This method uses stereo information.
 29.77 % 37.21 % 27.62 % 0.2 s 1 core @ 2.5 Ghz (Python + C/C++)
ERROR: Wrong syntax in BIBTEX file.
127 SparsePool code 27.92 % 35.52 % 25.87 % 0.13 s 8 cores @ 2.5 Ghz (Python)
Z. Wang, W. Zhan and M. Tomizuka: Fusing bird view lidar point cloud and front view camera image for deep object detection. arXiv preprint arXiv:1711.06703 2017.
128 AVOD
This method makes use of Velodyne laser scans.
 code 27.86 % 36.10 % 25.76 % 0.08 s Titan X (pascal)
J. Ku, M. Mozifian, J. Lee, A. Harakeh and S. Waslander: Joint 3D Proposal Generation and Object Detection from View Aggregation. IROS 2018.
129 CSW3D
This method makes use of Velodyne laser scans.
 26.64 % 33.75 % 23.34 % 0.03 s 4 cores @ 2.5 Ghz (C/C++)
J. Hu, T. Wu, H. Fu, Z. Wang and K. Ding: Cascaded Sliding Window Based Real-Time 3D Region Proposal for Pedestrian Detection. ROBIO 2019.
130 PointRGBNet 26.40 % 34.77 % 24.03 % 0.08 s 4 cores @ 2.5 Ghz (Python + C/C++)
131 Disp R-CNN (velo)
This method uses stereo information.
 code 25.80 % 37.12 % 22.04 % 0.387 s GPU @ 2.5 Ghz (Python + C/C++)
J. Sun, L. Chen, Y. Xie, S. Zhang, Q. Jiang, X. Zhou and H. Bao: Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation. CVPR 2020.
132 Disp R-CNN
This method uses stereo information.
 code 25.40 % 35.75 % 21.79 % 0.387 s GPU @ 2.5 Ghz (Python + C/C++)
J. Sun, L. Chen, Y. Xie, S. Zhang, Q. Jiang, X. Zhou and H. Bao: Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation. CVPR 2020.
133 FusionDetv2-v1 24.55 % 30.58 % 23.64 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
134 CG-Stereo
This method uses stereo information.
 24.31 % 33.22 % 20.95 % 0.57 s GeForce RTX 2080 Ti
C. Li, J. Ku and S. Waslander: Confidence Guided Stereo 3D Object Detection with Split Depth Estimation. IROS 2020.
135 LIGA-Stereo-old
This method uses stereo information.
 23.23 % 30.14 % 20.58 % 0.375 s Titan Xp
136 YOLOStereo3D
This method uses stereo information.
 code 19.75 % 28.49 % 16.48 % 0.1 s GPU 1080Ti
Y. Liu, L. Wang and M. Liu: YOLOStereo3D: A Step Back to 2D for Efficient Stereo 3D Detection. 2021 International Conference on Robotics and Automation (ICRA) 2021.
137 OSE+ 19.67 % 28.30 % 17.17 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
138 AEC3D 19.00 % 24.39 % 17.43 % 18 ms GPU @ 2.5 Ghz (Python)
139 BEVC 17.65 % 23.49 % 15.92 % 35ms GPU @ 1.5 Ghz (Python)
140 OC Stereo
This method uses stereo information.
 code 17.58 % 24.48 % 15.60 % 0.35 s 1 core @ 2.5 Ghz (Python + C/C++)
A. Pon, J. Ku, C. Li and S. Waslander: Object-Centric Stereo Matching for 3D Object Detection. ICRA 2020.
141 BirdNet
This method makes use of Velodyne laser scans.
 17.08 % 22.04 % 15.82 % 0.11 s Titan Xp (Caffe)
J. Beltrán, C. Guindel, F. Moreno, D. Cruzado, F. García and A. Escalera: BirdNet: A 3D Object Detection Framework from LiDAR Information. 2018 21st International Conference on Intelligent Transportation Systems (ITSC) 2018.
142 VN3D 15.69 % 19.56 % 13.17 % 0.02 s 1 core @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
143 DSGN
This method uses stereo information.
 code 15.55 % 20.53 % 14.15 % 0.67 s NVIDIA Tesla V100
Y. Chen, S. Liu, X. Shen and J. Jia: DSGN: Deep Stereo Geometry Network for 3D Object Detection. CVPR 2020.
144 SOD 14.68 % 21.13 % 12.67 % 0.1 s 1 core @ 2.5 Ghz (Python)
145 Complexer-YOLO
This method makes use of Velodyne laser scans.
 13.96 % 17.60 % 12.70 % 0.06 s GPU @ 3.5 Ghz (C/C++)
M. Simon, K. Amende, A. Kraus, J. Honer, T. Samann, H. Kaulbersch, S. Milz and H. Michael Gross: Complexer-YOLO: Real-Time 3D Object Detection and Tracking on Semantic Point Clouds. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops 2019.
146 RT3D-GMP
This method uses stereo information.
 11.41 % 16.23 % 10.12 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
H. Königshof and C. Stiller: Learning-Based Shape Estimation with Grid Map Patches for Realtime 3D Object Detection for Automated Driving. 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC) 2020.
147 DD3D code 11.04 % 16.64 % 9.38 % n/a s 1 core @ 2.5 Ghz (C/C++)
D. Park, R. Ambrus, V. Guizilini, J. Li and A. Gaidon: Is Pseudo-Lidar needed for Monocular 3D Object detection?. IEEE/CVF International Conference on Computer Vision (ICCV) .
148 PS-fld 10.82 % 16.95 % 9.26 % 0.25 s 1 core @ 2.5 Ghz (C/C++)
149 CMKD 10.39 % 16.89 % 9.29 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
150 EGFN
This method uses stereo information.
 10.27 % 14.05 % 9.02 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
151 MonoDTR 10.18 % 15.33 % 8.61 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
152 LT-M3OD 9.99 % 14.85 % 8.52 % 0.03 s 1 core @ 2.5 Ghz (Python)
153 GUPNet code 9.76 % 14.95 % 8.41 % NA s 1 core @ 2.5 Ghz (Python + C/C++)
Y. Lu, X. Ma, L. Yang, T. Zhang, Y. Liu, Q. Chu, J. Yan and W. Ouyang: Geometry Uncertainty Projection Network for Monocular 3D Object Detection. arXiv preprint arXiv:2107.13774 2021.
154 Lite-FPN-GUPNet 9.32 % 14.13 % 7.93 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
155 mono3d code 9.20 % 14.53 % 7.82 % TBD TBD
156 ZongmuMono3d code 9.18 % 14.23 % 7.82 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
157 SGM3D 8.81 % 13.99 % 7.26 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
Z. Zhou, L. Du, X. Ye, Z. Zou, X. Tan, E. Ding, L. Zhang, X. Xue and J. Feng: SGM3D: Stereo Guided Monocular 3D Object Detection. 2021.
158 SCSTSV-MonoFlex 8.75 % 13.10 % 7.38 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
159 gupnet_se 8.65 % 13.40 % 7.78 % 0.03s 1 core @ 2.5 Ghz (C/C++)
160 SwinMono3D 8.54 % 12.96 % 7.19 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
161 MonoCon code 8.41 % 13.10 % 6.94 % 0.02 s GPU @ 2.5 Ghz (Python)
T. Xianpeng Liu: Learning Auxiliary Monocular Contexts Helps Monocular 3D Object Detection. AAAI 2022.
162 HBD 8.33 % 13.47 % 6.99 % 0.12 s 1 core @ 2.5 Ghz (C/C++)
163 MonoEdge 8.33 % 12.11 % 7.03 % 0.05 s GPU @ 2.5 Ghz (Python)
164 GCDR 8.27 % 11.50 % 7.37 % 0.28 s 1 core @ 2.5 Ghz (Python)
165 MonoFlex 8.16 % 11.89 % 6.81 % 0.03 s 1 core @ 2.5 Ghz (Python)
166 CaDDN code 8.14 % 12.87 % 6.76 % 0.63 s GPU @ 2.5 Ghz (Python)
C. Reading, A. Harakeh, J. Chae and S. Waslander: Categorical Depth Distribution Network for Monocular 3D Object Detection. CVPR 2021.
167 MonoGround 7.89 % 12.37 % 7.13 % 0.03 s 1 core @ 2.5 Ghz (Python)
168 vadin-TBD 7.66 % 11.87 % 6.82 % 0.04 s 1 core @ 2.5 Ghz (Python)
169 MonoLCD 7.62 % 11.21 % 6.47 % 0.04 s 1 core @ 2.5 Ghz (Python)
170 K3D 7.60 % 12.58 % 6.73 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
171 ANM 7.54 % 11.92 % 6.37 % 0.02 s 1 core @ 2.5 Ghz (C/C++)
172 SAIC_ADC_Mono3D code 7.54 % 12.06 % 6.41 % 50 s GPU @ 2.5 Ghz (Python)
173 Mix-Teaching-M3D 7.47 % 11.67 % 6.61 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
174 LPCG-Monoflex 7.33 % 10.82 % 6.18 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
175 MonoDDE 7.32 % 11.13 % 6.67 % 0.04 s 1 core @ 2.5 Ghz (Python)
176 RefinedMPL 7.18 % 11.14 % 5.84 % 0.15 s GPU @ 2.5 Ghz (Python + C/C++)
J. Vianney, S. Aich and B. Liu: RefinedMPL: Refined Monocular PseudoLiDAR for 3D Object Detection in Autonomous Driving. arXiv preprint arXiv:1911.09712 2019.
177 MonoEdge-Rotate 7.02 % 10.47 % 5.84 % 0.05 s GPU @ 2.5 Ghz (Python)
178 TopNet-HighRes
This method makes use of Velodyne laser scans.
 6.92 % 10.40 % 6.63 % 101ms NVIDIA GeForce 1080 Ti (tensorflow-gpu)
S. Wirges, T. Fischer, C. Stiller and J. Frias: Object Detection and Classification in Occupancy Grid Maps Using Deep Convolutional Networks. 2018 21st International Conference on Intelligent Transportation Systems (ITSC) 2018.
179 MonoRUn code 6.78 % 10.88 % 5.83 % 0.07 s GPU @ 2.5 Ghz (Python + C/C++)
H. Chen, Y. Huang, W. Tian, Z. Gao and L. Xiong: MonoRUn: Monocular 3D Object Detection by Reconstruction and Uncertainty Propagation. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2021.
180 MonoPair 6.68 % 10.02 % 5.53 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Chen, L. Tai, K. Sun and M. Li: MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationships. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020.
181 mono3d 6.62 % 10.10 % 5.46 % 0.03 s GPU @ 2.5 Ghz (Python)
182 monodle code 6.55 % 9.64 % 5.44 % 0.04 s GPU @ 2.5 Ghz (Python)
X. Ma, Y. Zhang, D. Xu, D. Zhou, S. Yi, H. Li and W. Ouyang: Delving into Localization Errors for Monocular 3D Object Detection. CVPR 2021 .
183 MonoFlex 6.31 % 9.43 % 5.26 % 0.03 s GPU @ 2.5 Ghz (Python)
Y. Zhang, J. Lu and J. Zhou: Objects are Different: Flexible Monocular 3D Object Detection. CVPR 2021.
184 GAC3D++ 6.29 % 9.29 % 5.20 % 0.25 s 1 core @ 2.5 Ghz (C/C++)
185 RelationNet3D_dla34 code 6.22 % 9.28 % 5.23 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
186 E2E-DA 5.95 % 8.79 % 5.72 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
187 MK3D 5.72 % 9.19 % 4.54 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
188 MDNet 5.66 % 8.24 % 4.74 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
189 M3DSSD++ code 5.65 % 8.10 % 4.72 % 0.16s 1 core @ 2.5 Ghz (C/C++)
190 MonoGeo 5.63 % 8.00 % 4.71 % 0.05 s 1 core @ 2.5 Ghz (Python)
191 ICCV 5.25 % 8.34 % 4.72 % 0.04 s GPU @ 2.5 Ghz (Python)
192 MonoHMOO 5.23 % 7.62 % 4.28 % 0.2 s 1 core @ 2.5 Ghz (Python + C/C++)
193 RelationNet3D_res18 code 5.19 % 7.95 % 4.21 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
194 Aug3D-RPN 4.71 % 6.01 % 3.87 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
C. He, J. Huang, X. Hua and L. Zhang: Aug3D-RPN: Improving Monocular 3D Object Detection by Synthetic Images with Virtual Depth. 2021.
195 MM 4.70 % 7.81 % 4.01 % 1 s 1 core @ 2.5 Ghz (C/C++)
196 Shift R-CNN (mono) code 4.66 % 7.95 % 4.16 % 0.25 s GPU @ 1.5 Ghz (Python)
A. Naiden, V. Paunescu, G. Kim, B. Jeon and M. Leordeanu: Shift R-CNN: Deep Monocular 3D Object Detection With Closed-form Geometric Constraints. ICIP 2019.
197 Lite-FPN 4.38 % 6.57 % 3.56 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
198 COF3D 4.37 % 6.02 % 3.55 % 200 s 1 core @ 2.5 Ghz (Python + C/C++)
199 MAOLoss code 4.18 % 5.81 % 3.67 % 0.05 s 1 core @ 2.5 Ghz (Python)
200 MonoPSR code 4.00 % 6.12 % 3.30 % 0.2 s GPU @ 3.5 Ghz (Python)
J. Ku*, A. Pon* and S. Waslander: Monocular 3D Object Detection Leveraging Accurate Proposals and Shape Reconstruction. CVPR 2019.
201 MP-Mono 3.75 % 5.09 % 3.50 % 0.16 s GPU @ 2.5 Ghz (Python)
202 Geo3D 3.65 % 5.74 % 3.01 % 0.04 s GPU @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
203 DFR-Net 3.62 % 6.09 % 3.39 % 0.18 s 1080 Ti (Pytorch)
Z. Zou, X. Ye, L. Du, X. Cheng, X. Tan, L. Zhang, J. Feng, X. Xue and E. Ding: The devil is in the task: Exploiting reciprocal appearance-localization features for monocular 3d object detection . ICCV 2021.
204 DDMP-3D 3.55 % 4.93 % 3.01 % 0.18 s 1 core @ 2.5 Ghz (Python)
L. Wang, L. Du, X. Ye, Y. Fu, G. Guo, X. Xue, J. Feng and L. Zhang: Depth-conditioned Dynamic Message Propagation for Monocular 3D Object Detection. CVPR 2020.
205 E2E-DA-Lite (Res18) 3.51 % 5.82 % 3.42 % 0.01 s GPU @ 2.5 Ghz (Python)
206 M3D-RPN code 3.48 % 4.92 % 2.94 % 0.16 s GPU @ 1.5 Ghz (Python)
G. Brazil and X. Liu: M3D-RPN: Monocular 3D Region Proposal Network for Object Detection . ICCV 2019 .
207 D4LCN code 3.42 % 4.55 % 2.83 % 0.2 s GPU @ 2.5 Ghz (Python + C/C++)
M. Ding, Y. Huo, H. Yi, Z. Wang, J. Shi, Z. Lu and P. Luo: Learning Depth-Guided Convolutions for Monocular 3D Object Detection. CVPR 2020.
208 QD-3DT
This is an online method (no batch processing).
 code 3.37 % 5.53 % 3.02 % 0.03 s GPU @ 2.5 Ghz (Python)
H. Hu, Y. Yang, T. Fischer, F. Yu, T. Darrell and M. Sun: Monocular Quasi-Dense 3D Object Tracking. ArXiv:2103.07351 2021.
209 MonoEF code 2.79 % 4.27 % 2.21 % 0.03 s 1 core @ 2.5 Ghz (Python)
Y. Zhou, Y. He, H. Zhu, C. Wang, H. Li and Q. Jiang: Monocular 3D Object Detection: An Extrinsic Parameter Free Approach. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2021.
210 KAIST-VDCLab 2.47 % 3.27 % 2.43 % 0.04 s 1 core @ 2.5 Ghz (Python)
211 RT3DStereo
This method uses stereo information.
 2.45 % 3.28 % 2.35 % 0.08 s GPU @ 2.5 Ghz (C/C++)
H. Königshof, N. Salscheider and C. Stiller: Realtime 3D Object Detection for Automated Driving Using Stereo Vision and Semantic Information. Proc. IEEE Intl. Conf. Intelligent Transportation Systems 2019.
212 TopNet-UncEst
This method makes use of Velodyne laser scans.
 1.87 % 3.42 % 1.73 % 0.09 s NVIDIA GeForce 1080 Ti (tensorflow-gpu)
S. Wirges, M. Braun, M. Lauer and C. Stiller: Capturing Object Detection Uncertainty in Multi-Layer Grid Maps. 2019.
213 PPTrans 1.85 % 2.68 % 1.44 % 0.2 s GPU @ 2.5 Ghz (Python)
214 TBD 1.81 % 3.00 % 1.59 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
215 SS3D 1.78 % 2.31 % 1.48 % 48 ms Tesla V100 (Python)
E. Jörgensen, C. Zach and F. Kahl: Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss. CoRR 2019.
216 PGD-FCOS3D code 1.49 % 2.28 % 1.38 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
T. Wang, X. Zhu, J. Pang and D. Lin: Probabilistic and Geometric Depth: Detecting Objects in Perspective. Conference on Robot Learning (CoRL) 2021.
217 SparVox3D 1.35 % 1.93 % 1.04 % 0.05 s GPU @ 2.0 Ghz (Python)
E. Balatkan and F. Kıraç: Improving Regression Performance on Monocular 3D Object Detection Using Bin-Mixing and Sparse Voxel Data. 2021 6th International Conference on Computer Science and Engineering (UBMK) 2021.
218 EM code 1.18 % 1.09 % 0.80 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
219 CDTrack3D code 1.07 % 1.49 % 0.71 % 0.0106 s NVIDIA RTX 3090 GPU, i9 10850k CPU
220 EW code 0.81 % 0.79 % 0.74 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
221 mBoW
This method makes use of Velodyne laser scans.
 0.00 % 0.00 % 0.00 % 10 s 1 core @ 2.5 Ghz (C/C++)
J. Behley, V. Steinhage and A. Cremers: Laser-based Segment Classification Using a Mixture of Bag-of-Words. Proc. of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2013.
Table as LaTeX | Only published Methods

Cyclist


Method Setting Code Moderate Easy Hard Runtime Environment
1 CasA 73.47 % 87.91 % 66.17 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
2 ISE-RCNN-PV 71.94 % 84.94 % 64.09 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
3 SGNet 70.40 % 86.75 % 62.73 % 0.09 s GPU @ 2.5 Ghz (Python)
4 CAD 69.94 % 84.68 % 62.21 % 0.1 s GPU @ 2.5 Ghz (Python + C/C++)
5 SARFE 69.67 % 84.88 % 62.26 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
6 ISE-RCNN 69.18 % 82.62 % 62.77 % 0.09 s 1 core @ 2.5 Ghz (Python + C/C++)
7 anonymous code 69.13 % 83.09 % 61.35 % 0.05s 1 core @ >3.5 Ghz (python)
8 EQ-PVRCNN 69.10 % 85.41 % 62.30 % 0.2 s GPU @ 2.5 Ghz (Python + C/C++)
9 sa-voxel-centernet code 69.03 % 81.88 % 61.66 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
10 Fast VP-RCNN code 69.02 % 83.81 % 61.51 % 0.05 s 1 core @ 3.5 Ghz (C/C++)
11 SAA-PV-RCNN 68.96 % 82.06 % 61.54 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
12 CAT-Det 68.81 % 83.68 % 61.45 % 0.3 s GPU @ 2.5 Ghz (Python + C/C++)
13 BtcDet
This method makes use of Velodyne laser scans.
 code 68.68 % 82.81 % 61.81 % 0.09 s GPU @ 2.5 Ghz (Python + C/C++)
Q. Xu, Y. Zhong and U. Neumann: Behind the Curtain: Learning Occluded Shapes for 3D Object Detection. Proceedings of the AAAI Conference on Artificial Intelligence 2022.
14 SA-voxel-centernet code 68.67 % 81.47 % 61.40 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
15 DSA-PV-RCNN
This method makes use of Velodyne laser scans.
 code 68.54 % 82.19 % 61.33 % 0.08 s 1 core @ 2.5 Ghz (Python + C/C++)
P. Bhattacharyya, C. Huang and K. Czarnecki: SA-Det3D: Self-Attention Based Context-Aware 3D Object Detection. 2021.
16 FS-Net
This method makes use of Velodyne laser scans.
 68.35 % 81.81 % 60.96 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
17 TPCG 68.15 % 82.13 % 61.56 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
18 PE-RCVN 68.13 % 84.96 % 60.34 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
19 PDV 67.81 % 83.04 % 60.46 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
20 USVLab BSAODet (MM) 67.79 % 82.65 % 60.26 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
21 RangeIoUDet
This method makes use of Velodyne laser scans.
 67.77 % 83.12 % 60.26 % 0.02 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Liang, Z. Zhang, M. Zhang, X. Zhao and S. Pu: RangeIoUDet: Range Image Based Real-Time 3D Object Detector Optimized by Intersection Over Union. CVPR 2021.
22 Point Image Fusion 67.69 % 83.15 % 60.62 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
23 Generalized-SIENet 67.61 % 83.00 % 60.09 % 0.08 s 1 core @ 2.5 Ghz (Python)
24 FPC-RCNN 67.57 % 82.79 % 60.69 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
25 DDet 67.55 % 82.03 % 60.23 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
26 TBD
This method makes use of Velodyne laser scans.
 67.37 % 80.50 % 61.18 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
27 PV-RCNN++ 67.33 % 82.22 % 60.04 % 0.06 s 1 core @ 2.5 Ghz (C/C++)
28 USVLab BSAODet (SM) 67.25 % 81.94 % 59.19 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
29 SPG_mini
This method makes use of Velodyne laser scans.
 code 66.96 % 80.21 % 60.50 % 0.09 s GPU @ 2.5 Ghz (Python)
Q. Xu, Y. Zhou, W. Wang, C. Qi and D. Anguelov: SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation. Proceedings of the IEEE conference on computer vision and pattern recognition (ICCV) 2021.
30 VCRCNN 66.78 % 81.29 % 59.32 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
31 M3DeTR code 66.74 % 83.83 % 59.03 % n/a s GPU @ 1.0 Ghz (Python)
T. Guan, J. Wang, S. Lan, R. Chandra, Z. Wu, L. Davis and D. Manocha: M3DeTR: Multi-representation, Multi- scale, Mutual-relation 3D Object Detection with Transformers. 2021.
32 DSASNet 66.71 % 81.82 % 59.37 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
33 TBD 66.63 % 85.08 % 60.36 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
34 MSADet 66.49 % 84.21 % 59.73 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
35 VCT 66.38 % 82.37 % 60.01 % 0.2 s 1 core @ 2.5 Ghz (Python)
36 IA-SSD (single) 66.25 % 82.36 % 59.70 % 0.013 s 1 core @ 2.5 Ghz (C/C++)
37 EA-M-RCNN(BorderAtt) 66.04 % 82.39 % 58.19 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
38 WHUT-iou_ssd code 65.98 % 79.38 % 59.56 % 0.045s 1 core @ 2.5 Ghz (C/C++)
39 HotSpotNet 65.95 % 82.59 % 59.00 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
Q. Chen, L. Sun, Z. Wang, K. Jia and A. Yuille: object as hotspots. Proceedings of the European Conference on Computer Vision (ECCV) 2020.
40 ST-RCNN
This method makes use of Velodyne laser scans.
 65.61 % 78.82 % 58.44 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
41 TBD 65.48 % 79.90 % 57.95 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
42 JPVNet 65.41 % 80.66 % 59.26 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
43 Fast-CLOCs 65.31 % 82.83 % 57.43 % 0.1 s GPU @ 2.5 Ghz (Python)
S. Pang, D. Morris and H. Radha: Fast-CLOCs: Fast Camera-LiDAR Object Candidates Fusion for 3D Object Detection. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2022.
44 TCDVF 65.19 % 79.41 % 58.37 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
45 ASCNet 65.10 % 78.41 % 57.87 % 0.09 s 1 core @ 2.5 Ghz (C/C++)
46 F-ConvNet
This method makes use of Velodyne laser scans.
 code 65.07 % 81.98 % 56.54 % 0.47 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Wang and K. Jia: Frustum ConvNet: Sliding Frustums to Aggregate Local Point-Wise Features for Amodal 3D Object Detection. IROS 2019.
47 MVOD 64.95 % 79.52 % 57.53 % 0.16 s 1 core @ 2.5 Ghz (C/C++)
48 TBD 64.92 % 76.57 % 58.43 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
49 FPV-SSD 64.40 % 78.36 % 56.72 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
50 DGT-Det3D 64.38 % 78.27 % 57.79 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
51 AutoAlign 64.36 % 80.41 % 56.88 % 0.1 s 1 core @ 2.5 Ghz (Python)
52 FusionDetv2-v5 64.28 % 78.57 % 57.02 % 0.05 s 1 core @ 2.5 Ghz (Java + C/C++)
53 TBD 64.12 % 79.27 % 57.48 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
54 3DSSD code 64.10 % 82.48 % 56.90 % 0.04 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Yang, Y. Sun, S. Liu and J. Jia: 3DSSD: Point-based 3D Single Stage Object Detector. CVPR 2020.
55 VPFNet code 64.10 % 77.64 % 58.00 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
C. Wang, H. Chen and L. Fu: VPFNet: Voxel-Pixel Fusion Network for Multi-class 3D Object Detection. 2021.
56 PointPainting
This method makes use of Velodyne laser scans.
 63.78 % 77.63 % 55.89 % 0.4 s GPU @ 2.5 Ghz (Python + C/C++)
S. Vora, A. Lang, B. Helou and O. Beijbom: PointPainting: Sequential Fusion for 3D Object Detection. CVPR 2020.
57 MMLab PV-RCNN
This method makes use of Velodyne laser scans.
 code 63.71 % 78.60 % 57.65 % 0.08 s 1 core @ 2.5 Ghz (Python + C/C++)
S. Shi, C. Guo, L. Jiang, Z. Wang, J. Shi, X. Wang and H. Li: PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detection. CVPR 2020.
58 TBD_IOU 63.68 % 79.74 % 56.63 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
59 MMLab-PartA^2
This method makes use of Velodyne laser scans.
 code 63.52 % 79.17 % 56.93 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
S. Shi, Z. Wang, J. Shi, X. Wang and H. Li: From Points to Parts: 3D Object Detection from Point Cloud with Part-aware and Part-aggregation Network. IEEE Transactions on Pattern Analysis and Machine Intelligence 2020.
60 Point-GNN
This method makes use of Velodyne laser scans.
 code 63.48 % 78.60 % 57.08 % 0.6 s GPU @ 2.5 Ghz (Python)
W. Shi and R. Rajkumar: Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud. CVPR 2020.
61 MGAF-3DSSD code 63.43 % 80.64 % 55.15 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
J. Li, H. Dai, L. Shao and Y. Ding: Anchor-free 3D Single Stage Detector with Mask-Guided Attention for Point Cloud. MM '21: The 29th ACM International Conference on Multimedia (ACM MM) 2021.
62 FromVoxelToPoint code 63.41 % 81.49 % 56.40 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
J. Li, H. Dai, L. Shao and Y. Ding: From Voxel to Point: IoU-guided 3D Object Detection for Point Cloud with Voxel-to- Point Decoder. MM '21: The 29th ACM International Conference on Multimedia (ACM MM) 2021.
63 FusionDetv2-v4 63.38 % 79.65 % 56.61 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
64 P2V-RCNN 63.13 % 78.62 % 56.81 % 0.1 s 2 cores @ 2.5 Ghz (Python)
J. Li, S. Luo, Z. Zhu, H. Dai, A. Krylov, Y. Ding and L. Shao: P2V-RCNN: Point to Voxel Feature Learning for 3D Object Detection from Point Clouds. IEEE Access 2021.
65 H^23D R-CNN code 62.74 % 78.67 % 55.78 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
J. Deng, W. Zhou, Y. Zhang and H. Li: From Multi-View to Hollow-3D: Hallucinated Hollow-3D R-CNN for 3D Object Detection. IEEE Transactions on Circuits and Systems for Video Technology 2021.
66 TBD_IOU1 62.67 % 80.32 % 55.99 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
67 FPCR-CNN 62.56 % 79.61 % 55.82 % 0.05 s 1 core @ 2.5 Ghz (Python)
68 SVGA-Net 62.28 % 78.58 % 54.88 % 0.03s 1 core @ 2.5 Ghz (Python + C/C++)
Q. He, Z. Wang, H. Zeng, Y. Zeng and Y. Liu: SVGA-Net: Sparse Voxel-Graph Attention Network for 3D Object Detection from Point Clouds. AAAI 2022.
69 SRDL 62.02 % 77.35 % 55.52 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
ERROR: Wrong syntax in BIBTEX file.
70 FusionDetv1 62.02 % 77.33 % 55.52 % 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
71 Faraway-Frustum
This method makes use of Velodyne laser scans.
 code 62.00 % 77.36 % 55.40 % 0.1 s GPU @ 2.5 Ghz (Python)
H. Zhang, D. Yang, E. Yurtsever, K. Redmill and U. Ozguner: Faraway-Frustum: Dealing with Lidar Sparsity for 3D Object Detection using Fusion. arXiv preprint arXiv:2011.01404 2020.
72 DVFENet 62.00 % 78.73 % 55.18 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
Y. He, G. Xia, Y. Luo, L. Su, Z. Zhang, W. Li and P. Wang: DVFENet: Dual-branch Voxel Feature Extraction Network for 3D Object Detection. Neurocomputing 2021.
73 FusionDetv2-v3 61.96 % 79.43 % 55.28 % 0.05 s 1 core @ 2.5 Ghz (Python + C/C++)
74 NV2P-RCNN 61.95 % 73.58 % 55.62 % 0.1 s GPU @ 2.5 Ghz (Python)
75 IA-SSD (multi) 61.94 % 78.35 % 55.70 % 0.014 s 1 core @ 2.5 Ghz (C/C++)
76 VPN 61.82 % 77.81 % 55.33 % 0.06 s 1 core @ 2.5 Ghz (Python + C/C++)
77 FusionDetv2-v2 61.78 % 76.70 % 54.75 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
78 S-AT GCN 61.70 % 75.24 % 55.32 % 0.02 s GPU @ 2.0 Ghz (Python)
L. Wang, C. Wang, X. Zhang, T. Lan and J. Li: S-AT GCN: Spatial-Attention Graph Convolution Network based Feature Enhancement for 3D Object Detection. CoRR 2021.
79 SIF 61.61 % 77.13 % 55.11 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
P. An: SIF. Submitted to CVIU 2021.
80 STD code 61.59 % 78.69 % 55.30 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
Z. Yang, Y. Sun, S. Liu, X. Shen and J. Jia: STD: Sparse-to-Dense 3D Object Detector for Point Cloud. ICCV 2019.
81 Dune-DCF-e11 61.03 % 80.38 % 54.07 % 1 s 1 core @ 2.5 Ghz (C/C++)
82 SCIR-Net
This method makes use of Velodyne laser scans.
 60.89 % 76.32 % 54.48 % 0.03 s GPU @ 2.5 Ghz (Python + C/C++)
83 P2V_PCV1 60.84 % 75.25 % 54.70 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
84 City-CF 60.84 % 79.32 % 53.83 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
85 NV-RCNN 60.66 % 78.34 % 54.39 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
86 Dune-DCF-e15 60.53 % 78.68 % 53.73 % 1 s 1 core @ 2.5 Ghz (C/C++)
87 SAA-SECOND 60.50 % 75.65 % 53.81 % 38m s 1 core @ 2.5 Ghz (C/C++)
88 AB3DMOT
This method makes use of Velodyne laser scans.
This is an online method (no batch processing).
 code 60.30 % 75.42 % 53.81 % 0.0047s 1 core @ 2.5 Ghz (Python)
X. Weng and K. Kitani: A Baseline for 3D Multi-Object Tracking. arXiv:1907.03961 2019.
89 EPNet++ 59.71 % 76.15 % 53.67 % 0.1 s GPU @ 2.5 Ghz (Python)
90 TBD 59.61 % 74.98 % 53.52 % TBD GPU @ 2.5 Ghz (Python + C/C++)
91 City-CF-fixed 59.56 % 77.39 % 53.32 % 1 s 1 core @ 2.5 Ghz (C/C++)
92 XView 59.55 % 77.24 % 53.47 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
L. Xie, G. Xu, D. Cai and X. He: X-view: Non-egocentric Multi-View 3D Object Detector. 2021.
93 FPC3D_all
This method makes use of Velodyne laser scans.
 59.45 % 74.75 % 52.71 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
94 TANet code 59.44 % 75.70 % 52.53 % 0.035s GPU @ 2.5 Ghz (Python + C/C++)
Z. Liu, X. Zhao, T. Huang, R. Hu, Y. Zhou and X. Bai: TANet: Robust 3D Object Detection from Point Clouds with Triple Attention. AAAI 2020.
95 TBD_BD code 59.42 % 77.20 % 53.38 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
96 MMLab-PointRCNN
This method makes use of Velodyne laser scans.
 code 58.82 % 74.96 % 52.53 % 0.1 s GPU @ 2.5 Ghz (Python + C/C++)
S. Shi, X. Wang and H. Li: Pointrcnn: 3d object proposal generation and detection from point cloud. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2019.
97 CrazyTensor-CF 58.72 % 78.24 % 51.39 % 1 s 1 core @ 2.5 Ghz (C/C++)
98 HS3D code 58.65 % 74.75 % 52.98 % 0.12 s 1 core @ 2.5 Ghz (Python + C/C++)
99 PointPillars
This method makes use of Velodyne laser scans.
 code 58.65 % 77.10 % 51.92 % 16 ms 1080ti GPU and Intel i7 CPU
A. Lang, S. Vora, H. Caesar, L. Zhou, J. Yang and O. Beijbom: PointPillars: Fast Encoders for Object Detection from Point Clouds. CVPR 2019.
100 demo 58.20 % 72.48 % 52.69 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
101 ARPNET 58.20 % 74.21 % 52.13 % 0.08 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Ye, C. Zhang and X. Hao: ARPNET: attention region proposal network for 3D object detection. Science China Information Sciences 2019.
102 Dune-DCF-e09 57.82 % 74.49 % 51.53 % 1 s 1 core @ 2.5 Ghz (C/C++)
103 YF 56.86 % 71.37 % 52.18 % 0.04 s GPU @ 2.5 Ghz (C/C++)
104 LazyTorch-CP-Small-P 56.82 % 73.06 % 50.70 % 1 s 1 core @ 2.5 Ghz (C/C++)
105 LazyTorch-CP-Infer-O 56.77 % 73.03 % 50.73 % 1 s 1 core @ 2.5 Ghz (C/C++)
106 LazyTorch-CP 56.67 % 73.04 % 50.60 % 1 s 1 core @ 2.5 Ghz (C/C++)
107 FusionDetv2-baseline 56.34 % 71.16 % 50.70 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
108 Contrastive PP code 56.24 % 71.38 % 49.15 % 0.01 s 1 core @ 2.5 Ghz (Python)
109 epBRM
This method makes use of Velodyne laser scans.
 code 56.13 % 72.08 % 49.91 % 0.10 s 1 core @ 2.5 Ghz (C/C++)
K. Shin: Improving a Quality of 3D Object Detection by Spatial Transformation Mechanism. arXiv preprint arXiv:1910.04853 2019.
110 F-PointNet
This method makes use of Velodyne laser scans.
 code 56.12 % 72.27 % 49.01 % 0.17 s GPU @ 3.0 Ghz (Python)
C. Qi, W. Liu, C. Wu, H. Su and L. Guibas: Frustum PointNets for 3D Object Detection from RGB-D Data. arXiv preprint arXiv:1711.08488 2017.
111 CrazyTensor-CP 55.31 % 72.10 % 49.40 % 1 s 1 core @ 2.5 Ghz (Python)
112 PP-PCdet code 54.25 % 68.87 % 48.06 % 0.01 s 1 core @ 2.5 Ghz (Python)
113 TBD 53.95 % 70.44 % 47.22 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
114 TBD 53.95 % 70.44 % 47.22 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
115 BirdNet+
This method makes use of Velodyne laser scans.
 code 53.84 % 65.67 % 49.06 % 0.11 s Titan Xp (PyTorch)
A. Barrera, J. Beltrán, C. Guindel, J. Iglesias and F. García: BirdNet+: Two-Stage 3D Object Detection in LiDAR through a Sparsity-Invariant Bird’s Eye View. IEEE Access 2021.
116 tbd 53.00 % 68.71 % 46.32 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
117 PiFeNet 52.66 % 69.63 % 46.02 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
D. Le, H. Shi, H. Rezatofighi and J. Cai: PiFeNet: Pillar-Feature Network for Real-Time 3D Pedestrian Detection from Point Cloud. 2021.
118 PointRGBNet 52.15 % 67.05 % 46.78 % 0.08 s 4 cores @ 2.5 Ghz (Python + C/C++)
119 DisposalNet
This method uses stereo information.
 51.33 % 65.51 % 45.05 % 0.2 s 1 core @ 2.5 Ghz (Python + C/C++)
ERROR: Wrong syntax in BIBTEX file.
120 SCNet
This method makes use of Velodyne laser scans.
 50.79 % 67.98 % 45.15 % 0.04 s GPU @ 3.0 Ghz (Python)
Z. Wang, H. Fu, L. Wang, L. Xiao and B. Dai: SCNet: Subdivision Coding Network for Object Detection Based on 3D Point Cloud. IEEE Access 2019.
121 AVOD-FPN
This method makes use of Velodyne laser scans.
 code 50.55 % 63.76 % 44.93 % 0.1 s Titan X (Pascal)
J. Ku, M. Mozifian, J. Lee, A. Harakeh and S. Waslander: Joint 3D Proposal Generation and Object Detection from View Aggregation. IROS 2018.
122 MLOD
This method makes use of Velodyne laser scans.
 code 49.43 % 68.81 % 42.84 % 0.12 s GPU @ 1.5 Ghz (Python)
J. Deng and K. Czarnecki: MLOD: A multi-view 3D object detection based on robust feature fusion method. arXiv preprint arXiv:1909.04163 2019.
123 BirdNet+ (legacy)
This method makes use of Velodyne laser scans.
 code 47.72 % 67.38 % 42.89 % 0.1 s Titan Xp (PyTorch)
A. Barrera, C. Guindel, J. Beltrán and F. García: BirdNet+: End-to-End 3D Object Detection in LiDAR Bird’s Eye View. 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC) 2020.
124 PFF3D
This method makes use of Velodyne laser scans.
 code 46.78 % 63.27 % 41.37 % 0.05 s GPU @ 3.0 Ghz (Python + C/C++)
L. Wen and K. Jo: Fast and Accurate 3D Object Detection for Lidar-Camera-Based Autonomous Vehicles Using One Shared Voxel-Based Backbone. IEEE Access 2021.
125 DSGN++
This method uses stereo information.
 43.90 % 62.82 % 39.21 % 0.4 s NVIDIA Tesla V100
126 AVOD
This method makes use of Velodyne laser scans.
 code 42.08 % 57.19 % 38.29 % 0.08 s Titan X (pascal)
J. Ku, M. Mozifian, J. Lee, A. Harakeh and S. Waslander: Joint 3D Proposal Generation and Object Detection from View Aggregation. IROS 2018.
127 SparsePool code 37.33 % 52.61 % 33.39 % 0.13 s 8 cores @ 2.5 Ghz (Python)
Z. Wang, W. Zhan and M. Tomizuka: Fusing bird view lidar point cloud and front view camera image for deep object detection. arXiv preprint arXiv:1711.06703 2017.
128 LIGA-Stereo-old
This method uses stereo information.
 37.21 % 53.35 % 32.92 % 0.375 s Titan Xp
129 MMLAB LIGA-Stereo
This method uses stereo information.
 code 36.86 % 54.44 % 32.06 % 0.4 s 1 core @ 2.5 Ghz (Python + C/C++)
X. Guo, S. Shi, X. Wang and H. Li: LIGA-Stereo: Learning LiDAR Geometry Aware Representations for Stereo-based 3D Detector. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) 2021.
130 FusionDetv2-v1 36.58 % 51.38 % 32.88 % 0.04 s 1 core @ 2.5 Ghz (Python + C/C++)
131 SparsePool code 32.61 % 40.87 % 29.05 % 0.13 s 8 cores @ 2.5 Ghz (Python)
Z. Wang, W. Zhan and M. Tomizuka: Fusing bird view lidar point cloud and front view camera image for deep object detection. arXiv preprint arXiv:1711.06703 2017.
132 CG-Stereo
This method uses stereo information.
 30.89 % 47.40 % 27.23 % 0.57 s GeForce RTX 2080 Ti
C. Li, J. Ku and S. Waslander: Confidence Guided Stereo 3D Object Detection with Split Depth Estimation. IROS 2020.
133 BirdNet
This method makes use of Velodyne laser scans.
 30.25 % 43.98 % 27.21 % 0.11 s Titan Xp (Caffe)
J. Beltrán, C. Guindel, F. Moreno, D. Cruzado, F. García and A. Escalera: BirdNet: A 3D Object Detection Framework from LiDAR Information. 2018 21st International Conference on Intelligent Transportation Systems (ITSC) 2018.
134 SOD 25.29 % 40.51 % 21.32 % 0.1 s 1 core @ 2.5 Ghz (Python)
135 Disp R-CNN (velo)
This method uses stereo information.
 code 24.40 % 40.05 % 21.12 % 0.387 s GPU @ 2.5 Ghz (Python + C/C++)
J. Sun, L. Chen, Y. Xie, S. Zhang, Q. Jiang, X. Zhou and H. Bao: Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation. CVPR 2020.
136 Disp R-CNN
This method uses stereo information.
 code 24.40 % 40.04 % 21.12 % 0.387 s GPU @ 2.5 Ghz (Python + C/C++)
J. Sun, L. Chen, Y. Xie, S. Zhang, Q. Jiang, X. Zhou and H. Bao: Disp R-CNN: Stereo 3D Object Detection via Shape Prior Guided Instance Disparity Estimation. CVPR 2020.
137 AEC3D 22.41 % 31.40 % 21.56 % 18 ms GPU @ 2.5 Ghz (Python)
138 VN3D 21.53 % 30.76 % 21.03 % 0.02 s 1 core @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
139 OSE+ 20.75 % 32.62 % 17.23 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
140 Complexer-YOLO
This method makes use of Velodyne laser scans.
 18.53 % 24.27 % 17.31 % 0.06 s GPU @ 3.5 Ghz (C/C++)
M. Simon, K. Amende, A. Kraus, J. Honer, T. Samann, H. Kaulbersch, S. Milz and H. Michael Gross: Complexer-YOLO: Real-Time 3D Object Detection and Tracking on Semantic Point Clouds. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops 2019.
141 DSGN
This method uses stereo information.
 code 18.17 % 27.76 % 16.21 % 0.67 s NVIDIA Tesla V100
Y. Chen, S. Liu, X. Shen and J. Jia: DSGN: Deep Stereo Geometry Network for 3D Object Detection. CVPR 2020.
142 OC Stereo
This method uses stereo information.
 code 16.63 % 29.40 % 14.72 % 0.35 s 1 core @ 2.5 Ghz (Python + C/C++)
A. Pon, J. Ku, C. Li and S. Waslander: Object-Centric Stereo Matching for 3D Object Detection. ICRA 2020.
143 BEVC 14.08 % 22.30 % 13.44 % 35ms GPU @ 1.5 Ghz (Python)
144 RT3D-GMP
This method uses stereo information.
 12.99 % 18.31 % 10.63 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
H. Königshof and C. Stiller: Learning-Based Shape Estimation with Grid Map Patches for Realtime 3D Object Detection for Automated Driving. 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC) 2020.
145 EGFN
This method uses stereo information.
 7.69 % 13.84 % 6.75 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
146 PS-fld 6.18 % 11.22 % 5.21 % 0.25 s 1 core @ 2.5 Ghz (C/C++)
147 CMKD 5.68 % 9.27 % 4.99 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
148 mono3d code 5.27 % 10.08 % 4.12 % TBD TBD
149 CMKD 5.22 % 9.08 % 4.67 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
150 Mix-Teaching-M3D 4.91 % 8.04 % 4.15 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
151 DD3D code 4.79 % 7.52 % 4.22 % n/a s 1 core @ 2.5 Ghz (C/C++)
D. Park, R. Ambrus, V. Guizilini, J. Li and A. Gaidon: Is Pseudo-Lidar needed for Monocular 3D Object detection?. IEEE/CVF International Conference on Computer Vision (ICCV) .
152 MonoPSR code 4.74 % 8.37 % 3.68 % 0.2 s GPU @ 3.5 Ghz (Python)
J. Ku*, A. Pon* and S. Waslander: Monocular 3D Object Detection Leveraging Accurate Proposals and Shape Reconstruction. CVPR 2019.
153 TopNet-UncEst
This method makes use of Velodyne laser scans.
 4.54 % 7.13 % 3.81 % 0.09 s NVIDIA GeForce 1080 Ti (tensorflow-gpu)
S. Wirges, M. Braun, M. Lauer and C. Stiller: Capturing Object Detection Uncertainty in Multi-Layer Grid Maps. 2019.
154 LT-M3OD 4.52 % 7.87 % 4.38 % 0.03 s 1 core @ 2.5 Ghz (Python)
155 RelationNet3D_dla34 code 4.49 % 8.07 % 3.98 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
156 E2E-DA 4.42 % 7.36 % 3.34 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
157 LPCG-Monoflex 4.38 % 6.98 % 3.56 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
158 TBD 4.32 % 7.79 % 3.98 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
159 Lite-FPN-GUPNet 4.19 % 6.22 % 3.59 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
160 MAOLoss code 4.06 % 6.71 % 3.16 % 0.05 s 1 core @ 2.5 Ghz (Python)
161 E2E-DA-Lite (Res18) 3.99 % 6.87 % 3.04 % 0.01 s GPU @ 2.5 Ghz (Python)
162 MDNet 3.88 % 6.93 % 3.10 % 0.2 s 1 core @ 2.5 Ghz (C/C++)
163 SCSTSV-MonoFlex 3.82 % 6.65 % 3.06 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
164 SAIC_ADC_Mono3D code 3.81 % 6.73 % 3.03 % 50 s GPU @ 2.5 Ghz (Python)
165 MonoDDE 3.78 % 5.94 % 3.33 % 0.04 s 1 core @ 2.5 Ghz (Python)
166 ZongmuMono3d code 3.77 % 7.21 % 3.15 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
167 DFR-Net 3.58 % 5.69 % 3.10 % 0.18 s 1080 Ti (Pytorch)
Z. Zou, X. Ye, L. Du, X. Cheng, X. Tan, L. Zhang, J. Feng, X. Xue and E. Ding: The devil is in the task: Exploiting reciprocal appearance-localization features for monocular 3d object detection . ICCV 2021.
168 vadin-TBD 3.50 % 5.48 % 2.99 % 0.04 s 1 core @ 2.5 Ghz (Python)
169 CaDDN code 3.41 % 7.00 % 3.30 % 0.63 s GPU @ 2.5 Ghz (Python)
C. Reading, A. Harakeh, J. Chae and S. Waslander: Categorical Depth Distribution Network for Monocular 3D Object Detection. CVPR 2021.
170 RT3DStereo
This method uses stereo information.
 3.37 % 5.29 % 2.57 % 0.08 s GPU @ 2.5 Ghz (C/C++)
H. Königshof, N. Salscheider and C. Stiller: Realtime 3D Object Detection for Automated Driving Using Stereo Vision and Semantic Information. Proc. IEEE Intl. Conf. Intelligent Transportation Systems 2019.
171 MonoDTR 3.27 % 5.05 % 3.19 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
172 GUPNet code 3.21 % 5.58 % 2.66 % NA s 1 core @ 2.5 Ghz (Python + C/C++)
Y. Lu, X. Ma, L. Yang, T. Zhang, Y. Liu, Q. Chu, J. Yan and W. Ouyang: Geometry Uncertainty Projection Network for Monocular 3D Object Detection. arXiv preprint arXiv:2107.13774 2021.
173 M3DSSD++ code 2.94 % 5.18 % 2.43 % 0.16s 1 core @ 2.5 Ghz (C/C++)
174 MonoGeo 2.93 % 4.73 % 2.58 % 0.05 s 1 core @ 2.5 Ghz (Python)
175 SGM3D 2.92 % 5.49 % 2.64 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
Z. Zhou, L. Du, X. Ye, Z. Zou, X. Tan, E. Ding, L. Zhang, X. Xue and J. Feng: SGM3D: Stereo Guided Monocular 3D Object Detection. 2021.
176 K3D 2.81 % 5.17 % 2.57 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
177 MonoLCD 2.75 % 4.46 % 2.64 % 0.04 s 1 core @ 2.5 Ghz (Python)
178 ANM 2.69 % 4.69 % 2.68 % 0.02 s 1 core @ 2.5 Ghz (C/C++)
179 MonoGround 2.68 % 4.62 % 2.53 % 0.03 s 1 core @ 2.5 Ghz (Python)
180 monodle code 2.66 % 4.59 % 2.45 % 0.04 s GPU @ 2.5 Ghz (Python)
X. Ma, Y. Zhang, D. Xu, D. Zhou, S. Yi, H. Li and W. Ouyang: Delving into Localization Errors for Monocular 3D Object Detection. CVPR 2021 .
181 RelationNet3D_res18 code 2.55 % 4.85 % 2.33 % 0.04 s 1 core @ 2.5 Ghz (C/C++)
182 SwinMono3D 2.54 % 3.76 % 2.31 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
183 mono3d 2.53 % 4.71 % 2.22 % 0.03 s GPU @ 2.5 Ghz (Python)
184 MonoEdge-Rotate 2.51 % 4.28 % 2.13 % 0.05 s GPU @ 2.5 Ghz (Python)
185 DDMP-3D 2.50 % 4.18 % 2.32 % 0.18 s 1 core @ 2.5 Ghz (Python)
L. Wang, L. Du, X. Ye, Y. Fu, G. Guo, X. Xue, J. Feng and L. Zhang: Depth-conditioned Dynamic Message Propagation for Monocular 3D Object Detection. CVPR 2020.
186 MK3D 2.50 % 4.16 % 2.02 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
187 Aug3D-RPN 2.43 % 4.36 % 2.55 % 0.08 s 1 core @ 2.5 Ghz (C/C++)
C. He, J. Huang, X. Hua and L. Zhang: Aug3D-RPN: Improving Monocular 3D Object Detection by Synthetic Images with Virtual Depth. 2021.
188 QD-3DT
This is an online method (no batch processing).
 code 2.39 % 4.16 % 1.85 % 0.03 s GPU @ 2.5 Ghz (Python)
H. Hu, Y. Yang, T. Fischer, F. Yu, T. Darrell and M. Sun: Monocular Quasi-Dense 3D Object Tracking. ArXiv:2103.07351 2021.
189 MonoFlex 2.35 % 4.17 % 2.04 % 0.03 s GPU @ 2.5 Ghz (Python)
Y. Zhang, J. Lu and J. Zhou: Objects are Different: Flexible Monocular 3D Object Detection. CVPR 2021.
190 ICCV 2.33 % 4.51 % 2.22 % 0.04 s GPU @ 2.5 Ghz (Python)
191 MonoEdge 2.19 % 3.15 % 1.77 % 0.05 s GPU @ 2.5 Ghz (Python)
192 gupnet_se 2.13 % 3.84 % 2.13 % 0.03s 1 core @ 2.5 Ghz (C/C++)
193 MonoPair 2.12 % 3.79 % 1.83 % 0.06 s GPU @ 2.5 Ghz (Python + C/C++)
Y. Chen, L. Tai, K. Sun and M. Li: MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationships. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020.
194 GAC3D++ 2.12 % 3.82 % 2.23 % 0.25 s 1 core @ 2.5 Ghz (C/C++)
195 MonoFlex 2.10 % 3.39 % 1.67 % 0.03 s 1 core @ 2.5 Ghz (Python)
196 Geo3D 2.00 % 3.47 % 1.52 % 0.04 s GPU @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
197 MonoCon code 1.92 % 2.80 % 1.55 % 0.02 s GPU @ 2.5 Ghz (Python)
T. Xianpeng Liu: Learning Auxiliary Monocular Contexts Helps Monocular 3D Object Detection. AAAI 2022.
198 RefinedMPL 1.82 % 3.23 % 1.77 % 0.15 s GPU @ 2.5 Ghz (Python + C/C++)
J. Vianney, S. Aich and B. Liu: RefinedMPL: Refined Monocular PseudoLiDAR for 3D Object Detection in Autonomous Driving. arXiv preprint arXiv:1911.09712 2019.
199 TopNet-HighRes
This method makes use of Velodyne laser scans.
 1.67 % 2.49 % 1.88 % 101ms NVIDIA GeForce 1080 Ti (tensorflow-gpu)
S. Wirges, T. Fischer, C. Stiller and J. Frias: Object Detection and Classification in Occupancy Grid Maps Using Deep Convolutional Networks. 2018 21st International Conference on Intelligent Transportation Systems (ITSC) 2018.
200 D4LCN code 1.67 % 2.45 % 1.36 % 0.2 s GPU @ 2.5 Ghz (Python + C/C++)
M. Ding, Y. Huo, H. Yi, Z. Wang, J. Shi, Z. Lu and P. Luo: Learning Depth-Guided Convolutions for Monocular 3D Object Detection. CVPR 2020.
201 MonoHMOO 1.60 % 1.87 % 1.66 % 0.2 s 1 core @ 2.5 Ghz (Python + C/C++)
202 MP-Mono 1.58 % 2.36 % 1.69 % 0.16 s GPU @ 2.5 Ghz (Python)
203 KAIST-VDCLab 1.56 % 2.34 % 1.61 % 0.04 s 1 core @ 2.5 Ghz (Python)
204 COF3D 1.46 % 2.34 % 1.28 % 200 s 1 core @ 2.5 Ghz (Python + C/C++)
205 SS3D 1.45 % 2.80 % 1.35 % 48 ms Tesla V100 (Python)
E. Jörgensen, C. Zach and F. Kahl: Monocular 3D Object Detection and Box Fitting Trained End-to-End Using Intersection-over-Union Loss. CoRR 2019.
206 PGD-FCOS3D code 1.38 % 2.81 % 1.20 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
T. Wang, X. Zhu, J. Pang and D. Lin: Probabilistic and Geometric Depth: Detecting Objects in Perspective. Conference on Robot Learning (CoRL) 2021.
207 PPTrans 1.38 % 2.31 % 1.20 % 0.2 s GPU @ 2.5 Ghz (Python)
208 HBD 1.24 % 2.45 % 1.22 % 0.12 s 1 core @ 2.5 Ghz (C/C++)
209 GCDR 1.17 % 1.96 % 1.02 % 0.28 s 1 core @ 2.5 Ghz (Python)
210 MonoEF code 0.92 % 1.80 % 0.71 % 0.03 s 1 core @ 2.5 Ghz (Python)
Y. Zhou, Y. He, H. Zhu, C. Wang, H. Li and Q. Jiang: Monocular 3D Object Detection: An Extrinsic Parameter Free Approach. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2021.
211 M3D-RPN code 0.65 % 0.94 % 0.47 % 0.16 s GPU @ 1.5 Ghz (Python)
G. Brazil and X. Liu: M3D-RPN: Monocular 3D Region Proposal Network for Object Detection . ICCV 2019 .
212 MonoRUn code 0.61 % 1.01 % 0.48 % 0.07 s GPU @ 2.5 Ghz (Python + C/C++)
H. Chen, Y. Huang, W. Tian, Z. Gao and L. Xiong: MonoRUn: Monocular 3D Object Detection by Reconstruction and Uncertainty Propagation. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2021.
213 Lite-FPN 0.41 % 0.50 % 0.24 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
214 Shift R-CNN (mono) code 0.29 % 0.48 % 0.31 % 0.25 s GPU @ 1.5 Ghz (Python)
A. Naiden, V. Paunescu, G. Kim, B. Jeon and M. Leordeanu: Shift R-CNN: Deep Monocular 3D Object Detection With Closed-form Geometric Constraints. ICIP 2019.
215 MM 0.27 % 0.48 % 0.24 % 1 s 1 core @ 2.5 Ghz (C/C++)
216 CDTrack3D code 0.06 % 0.06 % 0.05 % 0.0106 s NVIDIA RTX 3090 GPU, i9 10850k CPU
217 mBoW
This method makes use of Velodyne laser scans.
 0.00 % 0.00 % 0.00 % 10 s 1 core @ 2.5 Ghz (C/C++)
J. Behley, V. Steinhage and A. Cremers: Laser-based Segment Classification Using a Mixture of Bag-of-Words. Proc. of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2013.
Table as LaTeX | Only published Methods

Related Datasets

Citation

When using this dataset in your research, we will be happy if you cite us:
@INPROCEEDINGS{Geiger2012CVPR,
  author = {Andreas Geiger and Philip Lenz and Raquel Urtasun},
  title = {Are we ready for Autonomous Driving? The KITTI Vision Benchmark Suite},
  booktitle = {Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2012}
}



eXTReMe Tracker