Andreas Geiger (MPI Tübingen) | Philip Lenz (KIT) | Christoph Stiller (KIT) | Raquel Urtasun (University of Toronto)

Stereo Evaluation 2015


The stereo 2015 / flow 2015 / scene flow 2015 benchmark consists of 200 training scenes and 200 test scenes (4 color images per scene, saved in loss less png format). Compared to the stereo 2012 and flow 2012 benchmarks, it comprises dynamic scenes for which the ground truth has been established in a semi-automatic process. Our evaluation server computes the percentage of bad pixels averaged over all ground truth pixels of all 200 test images. For this benchmark, we consider a pixel to be correctly estimated if the disparity or flow end-point error is <3px or <5% (for scene flow this criterion needs to be fulfilled for both disparity maps and the flow map). 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 disparity maps and flow fields. More details can be found in Object Scene Flow for Autonomous Vehicles (CVPR 2015).

Our evaluation table ranks all methods according to the number of erroneous pixels. All methods providing less than 100 % density have been interpolated using simple background interpolation as explained in the corresponding header file in the development kit. Legend:

  • D1: Percentage of stereo disparity outliers in first frame
  • D2: Percentage of stereo disparity outliers in second frame
  • Fl: Percentage of optical flow outliers
  • SF: Percentage of scene flow outliers (=outliers in either D0, D1 or Fl)
  • bg: Percentage of outliers averaged only over background regions
  • fg: Percentage of outliers averaged only over foreground regions
  • all: Percentage of outliers averaged over all ground truth pixels


Note: On 13.03.2017 we have fixed several small errors in the flow (noc+occ) ground truth of the dynamic foreground objects and manually verified all images for correctness by warping them according to the ground truth. As a consequence, all error numbers have decreased slightly. Please download the devkit and the annotations with the improved ground truth for the training set again if you have downloaded the files prior to 13.03.2017 and consider reporting these new number in all future publications. The last leaderboards before these corrections can be found here (optical flow 2015) and here (scene flow 2015). The leaderboards for the KITTI 2015 stereo benchmarks did not change.

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
  • Flow: Method uses optical flow (2 temporally adjacent images)
  • Multiview: Method uses more than 2 temporally adjacent images
  • Motion stereo: Method uses epipolar geometry for computing optical flow
  • Additional training data: Use of additional data sources for training (see details)

Evaluation ground truth        Evaluation area

Method Setting Code D1-bg D1-fg D1-all Density Runtime Environment
1 LEAStereo 1.40 % 2.91 % 1.65 % 100.00 % 0.30 s GPU @ 2.5 Ghz (Python)
2 MSMD-Net(only MS) 1.41 % 3.13 % 1.69 % 100.00 % 0.32 s 1 core @ 2.5 Ghz (C/C++)
3 Dahua_SF
This method uses optical flow information.
 1.48 % 2.83 % 1.71 % 100.00 % 0.5 s 1 core @ 2.5 Ghz (Python)
4 Dahua_Stereo 1.48 % 2.83 % 1.71 % 100.00 % 1.52 s GPU @ 2.5 Ghz (Python)
5 CSPN 1.51 % 2.88 % 1.74 % 100.00 % 1.0 s GPU @ 2.5 Ghz (Python)
X. Cheng, P. Wang and R. Yang: Learning Depth with Convolutional Spatial Propagation Network. IEEE Transactions on Pattern Analysis and Machine Intelligence(T-PAMI) 2019.
6 DJIStereo 1.46 % 3.20 % 1.75 % 100.00 % 1.5 s 1 core @ 2.5 Ghz (Python)
7 HorizonStereo 1.46 % 3.26 % 1.76 % 100.00 % 1.8 s GPU@ 2.5 Ghz (Python)
8 DHSM 1.54 % 2.92 % 1.77 % 100.00 % 2 s 1 core @ 2.5 Ghz (Python)
9 NLCA-Net_V2 1.41 % 3.56 % 1.77 % 100.00 % 0.67 s 1 core @ 2.5 Ghz (C/C++)
10 DSMNet+GANet 1.48 % 3.23 % 1.77 % 100.00 % 1.9 s 1 core @ 2.5 Ghz (Python)
11 GANet++ 1.55 % 2.96 % 1.78 % 100.00 % 1.52 s GPU @ 2.5 Ghz (Python)
12 SUW-Stereo 1.47 % 3.45 % 1.80 % 100.00 % 1.8 s 1 core @ 2.5 Ghz (C/C++)
H. Ren, A. Raj, M. El-Khamy and J. Lee: SUW-Learn: Joint Supervised, Unsupervised, Weakly Supervised Deep Learning for Monocular Depth Estimation. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops 2020.
13 PVStereo 1.48 % 3.46 % 1.81 % 100.00 % 1.8 s GPU @ 2.5 Ghz (Python)
14 GANet-deep code 1.48 % 3.46 % 1.81 % 100.00 % 1.8 s GPU @ 2.5 Ghz (Python)
F. Zhang, V. Prisacariu, R. Yang and P. Torr: GA-Net: Guided Aggregation Net for End-to-end Stereo Matching. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
15 StereoExp-v2
This method uses optical flow information.
 1.48 % 3.46 % 1.81 % 100.00 % 2 s GPU @ 2.5 Ghz (Python)
16 Stereo expansion
This method uses optical flow information.
 code 1.48 % 3.46 % 1.81 % 100.00 % 2 s GPU @ 2.5 Ghz (Python)
G. Yang and D. Ramanan: Upgrading Optical Flow to 3D Scene Flow through Optical Expansion. CVPR 2020.
17 GA-Net+G 1.49 % 3.47 % 1.82 % 100.00 % 0.5 s GPU (Python)
18 PSG-GANet 1.49 % 3.46 % 1.82 % 100.00 % 1.8 s GPU @ 2.5 Ghz (Python)
19 HDU-FCC 1.50 % 3.45 % 1.82 % 100.00 % 0.70 s 1 core @ 2.5 Ghz (C/C++)
20 CMF 1.44 % 3.76 % 1.83 % 100.00 % 0.5 s GPU @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
21 NLCA-Net-3 code 1.45 % 3.78 % 1.83 % 100.00 % 0.44 s >8 cores @ 3.5 Ghz (C/C++)
Z. Rao, M. He, Y. Dai, Z. Zhu, B. Li and R. He: NLCA-Net: a non-local context attention network for stereo matching. APSIPA Transactions on Signal and Information Processing 2020.
22 HD3+_Flow
This method uses optical flow information.
 1.63 % 2.89 % 1.84 % 100.00 % 0.04 s GPU @ 2.5 Ghz (Python)
23 UnDAF-GANet 1.51 % 3.51 % 1.84 % 100.00 % 1.8 s GPU @ 2.5 Ghz (Python)
24 AMNet 1.53 % 3.43 % 1.84 % 100.00 % 0.9 s GPU @ 2.5 Ghz (Python)
X. Du, M. El-Khamy and J. Lee: AMNet: Deep Atrous Multiscale Stereo Disparity Estimation Networks. 2019.
25 CVF-BPNet 1.57 % 3.29 % 1.85 % 100.00 % 0.75 s GPU @ NVIDIA GTX-1080Ti (Pytorch)
26 NVstereo3D 1.52 % 3.54 % 1.86 % 100.00 % 0.15 s GPU @ 2.5 Ghz (Python)
27 MFM-Net 1.51 % 3.67 % 1.87 % 100.00 % 0.47 s GPU @ 1.5 Ghz (Python)
28 RAS-Net 1.61 % 3.16 % 1.87 % 100.00 % 0.6 s 1 core @ 2.5 Ghz (C/C++)
29 AcfNet code 1.51 % 3.80 % 1.89 % 100.00 % 0.48 s GPU @ 2.5 Ghz (Python)
Y. Zhang, Y. Chen, X. Bai, S. Yu, K. Yu, Z. Li and K. Yang: Adaptive Unimodal Cost Volume Filtering for Deep Stereo Matching. AAAI 2020.
30 LISAStereo 1.62 % 3.24 % 1.89 % 100.00 % 0.09 s 4 cores @ 2.5 Ghz (Python)
31 DSMNet-finetune 1.65 % 3.16 % 1.90 % 100.00 % 1.5 s GPU @ 2.5 Ghz (Python)
32 CAIS+PSMNet 1.57 % 3.62 % 1.91 % 100.00 % 0.38 s GPU @ 2.5 Ghz (Python)
33 GAN 1.62 % 3.34 % 1.91 % 100.00 % 1.8 s GPU @ 2.5 Ghz (Python)
34 NLCA_NET_v2_RVC 1.51 % 3.97 % 1.92 % 100.00 % 0.67 s GPU @ 2.5 Ghz (Python)
Z. Rao, M. He, Y. Dai, Z. Zhu, B. Li and R. He: NLCA-Net: a non-local context attention network for stereo matching. APSIPA Transactions on Signal and Information Processing 2020.
35 CDN code 1.66 % 3.20 % 1.92 % 100.00 % 0.4 s GPU @ 2.5 Ghz (Python)
36 HCGANet 1.64 % 3.38 % 1.93 % 100.00 % 0.064 s GPU @ 2.5 Ghz (Python)
37 GANet-15 code 1.55 % 3.82 % 1.93 % 100.00 % 0.36 s GPU (Pytorch)
F. Zhang, V. Prisacariu, R. Yang and P. Torr: GA-Net: Guided Aggregation Net for End-to-end Stereo Matching. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
38 NLCA-Net code 1.53 % 4.09 % 1.96 % 100.00 % 0.6 s 1 core @ 2.5 Ghz (C/C++)
Z. Rao, M. He, Y. Dai, Z. Zhu, B. Li and R. He: NLCA-Net: a non-local context attention network for stereo matching. APSIPA Transactions on Signal and Information Processing 2020.
39 CFNet_RVC 1.65 % 3.53 % 1.96 % 100.00 % 0.22 s GPU @ 2.5 Ghz (Python)
40 FASM 1.69 % 3.37 % 1.97 % 100.00 % 0.42 s 1 core @ 2.5 Ghz (Python)
41 MonoStereo 1.63 % 3.73 % 1.98 % 100.00 % 0.05 s GPU @ 2.5 Ghz (Python)
42 DPCTF-S 1.71 % 3.34 % 1.98 % 100.00 % 0.11 s GPU @ 2.5 Ghz (Python)
43 WTHNet 1.63 % 3.75 % 1.98 % 100.00 % 0.5 s 1 core @ 2.5 Ghz (C/C++)
44 HITNet code 1.74 % 3.20 % 1.98 % 100.00 % 0.015 s Titan V,
V. Tankovich, C. Häne, S. Fanello, Y. Zhang, S. Izadi and S. Bouaziz: HITNet: Hierarchical Iterative Tile Refinement Network for Real-time Stereo Matching. 2020.
45 TS_FAD 1.85 % 2.69 % 1.99 % 100.00 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
46 SGNet 1.63 % 3.76 % 1.99 % 100.00 % 0.6 s 1 core @ 2.5 Ghz (Python + C/C++)
47 CSN code 1.59 % 4.03 % 2.00 % 100.00 % 0.6 s 1 core @ 2.5 Ghz (Python)
X. Gu, Z. Fan, S. Zhu, Z. Dai, F. Tan and P. Tan: Cascade cost volume for high-resolution multi-view stereo and stereo matching. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020.
48 GwcNet-cmd 1.57 % 4.14 % 2.00 % 100.00 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
49 MANet-Selected 1.58 % 4.13 % 2.00 % 100.00 % 0.88 s GPU @ 2.5 Ghz (Python)
50 DHSM 1.76 % 3.33 % 2.02 % 100.00 % 2 s 1 core @ 2.5 Ghz (Python)
51 HD^3-Stereo code 1.70 % 3.63 % 2.02 % 100.00 % 0.14 s NVIDIA Pascal Titan XP
Z. Yin, T. Darrell and F. Yu: Hierarchical Discrete Distribution Decomposition for Match Density Estimation. CVPR 2019.
52 CVL 1.71 % 3.59 % 2.02 % 100.00 % 0.36 s 1 core @ 2.5 Ghz (Python)
53 AANet+ code 1.65 % 3.96 % 2.03 % 100.00 % 0.06 s NVIDIA V100 GPU
H. Xu and J. Zhang: AANet: Adaptive Aggregation Network for Efficient Stereo Matching. CVPR 2020.
54 GwcNet_CSA 1.73 % 3.57 % 2.03 % 100.00 % 0.37 s 1 core @ 2.5 Ghz (C/C++)
55 DHSM 1.78 % 3.34 % 2.04 % 100.00 % 1.9 s 1 core @ 2.5 Ghz (Python)
56 MANet-Medium 1.63 % 4.15 % 2.05 % 100.00 % 0.88 s GPU @ 2.5 Ghz (Python)
57 MANet-Selected 1.61 % 4.29 % 2.06 % 100.00 % 0.88 s GPU @ 2.5 Ghz (Python)
58 LR-PSMNet 1.65 % 4.13 % 2.06 % 100.00 % 0.5 s GPU @ 2.5 Ghz (Python)
59 PSMNet++ 1.63 % 4.27 % 2.07 % 100.00 % 0.36 s GPU @ >3.5 Ghz (Python)
60 ISSGA-Net 1.73 % 3.78 % 2.07 % 100.00 % 0.38 s GPU @ 2.5 Ghz (Python)
61 DeepStereo 1.71 % 3.87 % 2.07 % 100.00 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
62 UnDAF-SENSE
This method uses optical flow information.
 1.75 % 3.70 % 2.07 % 100.00 % 0.32 s GPU @ 2.5 Ghz (Python)
63 MANet-Large 1.61 % 4.41 % 2.08 % 100.00 % 0.88 s GPU @ 2.5 Ghz (Python)
64 Gwc-MSDRNet code 1.73 % 3.84 % 2.08 % 100.00 % 0.25 s 1 core @ 2.5 Ghz (Python)
65 MGSNet 1.68 % 4.06 % 2.08 % 100.00 % 0.65 s GPU @ 2.5 Ghz (Python)
66 EdgeStereo-V2 1.84 % 3.30 % 2.08 % 100.00 % 0.32s Nvidia GTX Titan Xp
X. Song, X. Zhao, L. Fang, H. Hu and Y. Yu: Edgestereo: An effective multi-task learning network for stereo matching and edge detection. International Journal of Computer Vision (IJCV) 2019.
67 Gwc-MSRef 1.73 % 3.87 % 2.08 % 100.00 % 0.2 s GPU @ 2.5 Ghz (Python)
68 DDP_in_HD3 1.78 % 3.64 % 2.09 % 100.00 % 1 s 1 core @ 2.5 Ghz (Python + C/C++)
69 DDP_out_HD3 1.78 % 3.66 % 2.09 % 100.00 % 10 min 1 GPU (Python)
70 MDA-Net(New) 1.76 % 3.77 % 2.10 % 100.00 % 0.4 s 1 core @ 2.5 Ghz (Python)
71 PSMNet_CSA 1.76 % 3.79 % 2.10 % 100.00 % 0.47 s 1 core @ 2.5 Ghz (Python)
72 GwcNet-g code 1.74 % 3.93 % 2.11 % 100.00 % 0.32 s GPU @ 2.0 Ghz (Python + C/C++)
X. Guo, K. Yang, W. Yang, X. Wang and H. Li: Group-wise correlation stereo network. CVPR 2019.
73 False 1.75 % 3.93 % 2.11 % 100.00 % 0.4 s 1 core @ 2.5 Ghz (Python)
74 SPOSF 1.77 % 3.81 % 2.11 % 100.00 % 10 min 1 core @ 3.5 Ghz (Matlab + C/C++)
75 SSPCVNet 1.75 % 3.89 % 2.11 % 100.00 % 0.9 s 1 core @ 2.5 Ghz (Python)
Z. Wu, X. Wu, X. Zhang, S. Wang and L. Ju: Semantic Stereo Matching With Pyramid Cost Volumes. The IEEE International Conference on Computer Vision (ICCV) 2019.
76 DHSM_atte 1.79 % 3.74 % 2.11 % 100.00 % 1.9 s 1 core @ 2.5 Ghz (Python)
77 SDEA 1.71 % 4.17 % 2.12 % 100.00 % 0.40 s 1 core @ 2.5 Ghz (Python)
78 WSMCnet code 1.72 % 4.19 % 2.13 % 100.00 % 0.39s Nvidia GTX 1070 (Pytorch)
Y. Wang, H. Wang, G. Yu, M. Yang, Y. Yuan and J. Quan: Stereo Matching Algorithm Based on Three-Dimensional Convolutional Neural Network. Acta Optica Sinica 2019.
79 DEINet+ft 1.72 % 4.26 % 2.14 % 100.00 % 0.23 s GPU @ 2.5 Ghz (Python + C/C++)
80 HSM-1.8x code 1.80 % 3.85 % 2.14 % 100.00 % 0.14 s Titan X Pascal
G. Yang, J. Manela, M. Happold and D. Ramanan: Hierarchical Deep Stereo Matching on High- Resolution Images. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
81 DSMNet 1.78 % 3.97 % 2.14 % 100.00 % 0.67 s 1 core @ 2.5 Ghz (Python)
82 DFNet 1.78 % 4.03 % 2.15 % 100.00 % 0.7 s 1 core @ 2.5 Ghz (Python)
83 DeepPruner (best) code 1.87 % 3.56 % 2.15 % 100.00 % 0.18 s 1 core @ 2.5 Ghz (C/C++)
S. Duggal, S. Wang, W. Ma, R. Hu and R. Urtasun: DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch. ICCV 2019.
84 GANet++ 1.88 % 3.53 % 2.16 % 100.00 % 0.04 s GPU @ 2.5 Ghz (Python)
85 PSM+CRF 1.86 % 3.66 % 2.16 % 100.00 % 0.32 s GPU @ 2.0 Ghz (C/C++)
86 Stereo-fusion-SJTU 1.87 % 3.61 % 2.16 % 100.00 % 0.7 s Nvidia GTX Titan Xp
X. Song, X. Zhao, H. Hu and L. Fang: EdgeStereo: A Context Integrated Residual Pyramid Network for Stereo Matching. Asian Conference on Computer Vision 2018.
87 AutoDispNet-CSS code 1.94 % 3.37 % 2.18 % 100.00 % 0.9 s 1 core @ 2.5 Ghz (C/C++)
T. Saikia, Y. Marrakchi, A. Zela, F. Hutter and T. Brox: AutoDispNet: Improving Disparity Estimation with AutoML. The IEEE International Conference on Computer Vision (ICCV) 2019.
88 MAN 1.74 % 4.44 % 2.19 % 100.00 % 1.65 s 1 core @ 2.5 Ghz (Python)
89 Bi3D code 1.95 % 3.48 % 2.21 % 100.00 % 0.48 s GPU @ 1.5 Ghz (Python)
A. Badki, A. Troccoli, K. Kim, J. Kautz, P. Sen and O. Gallo: Bi3D: Stereo Depth Estimation via Binary Classifications. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020.
90 MFRANet 1.78 % 4.35 % 2.21 % 100.00 % 0.32 s GPU @ 2.5 Ghz (Python)
91 ICANet 1.81 % 4.23 % 2.21 % 100.00 % 0.47 s GPU @ 2.5 Ghz (Python)
92 dh 1.86 % 4.01 % 2.22 % 100.00 % 1.9 s 1 core @ 2.5 Ghz (C/C++)
F. Zhang, V. Prisacariu, R. Yang and P. Torr: GA-Net: Guided Aggregation Net for End-to-end Stereo Matching. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
93 MPA-Net 1.78 % 4.43 % 2.22 % 100.00 % 0.86 s 1 core @ 2.5 Ghz (Python)
94 SENSE
This method uses optical flow information.
 code 2.07 % 3.01 % 2.22 % 100.00 % 0.32s GPU, GTX 2080Ti
H. Jiang, D. Sun, V. Jampani, Z. Lv, E. Learned-Miller and J. Kautz: SENSE: A Shared Encoder Network for Scene-Flow Estimation. The IEEE International Conference on Computer Vision (ICCV) 2019.
95 CTFNet-v2 1.80 % 4.46 % 2.24 % 100.00 % 0.7 s 8 cores @ 2.5 Ghz (Python)
96 SegStereo code 1.88 % 4.07 % 2.25 % 100.00 % 0.6 s Nvidia GTX Titan Xp
G. Yang, H. Zhao, J. Shi, Z. Deng and J. Jia: SegStereo: Exploiting Semantic Information for Disparity Estimation. ECCV 2018.
97 PSMNet+GLR 1.85 % 4.25 % 2.25 % 100.00 % 0.3 s GPU (Python)
98 NWS 1.88 % 4.13 % 2.26 % 100.00 % 0.41 s 1 core @ 2.5 Ghz (Python)
99 CTFNet 1.81 % 4.56 % 2.27 % 100.00 % 0.7 s 8 cores @ 2.5 Ghz (Python)
100 MCV-MFC 1.95 % 3.84 % 2.27 % 100.00 % 0.35 s 1 core @ 2.5 Ghz (C/C++)
Z. Liang, Y. Guo, Y. Feng, W. Chen, L. Qiao, L. Zhou, J. Zhang and H. Liu: Stereo Matching Using Multi-level Cost Volume and Multi-scale Feature Constancy. IEEE transactions on pattern analysis and machine intelligence 2019.
101 CTFNet-v1 1.85 % 4.35 % 2.27 % 100.00 % 0.6 s 8 cores @ 2.5 Ghz (Python)
102 NS-PSM 1.86 % 4.35 % 2.27 % 100.00 % 0.5 s 1 core @ 2.5 Ghz (Python)
103 HSM-1.5x code 1.95 % 3.93 % 2.28 % 100.00 % 0.085 s Titan X Pascal
G. Yang, J. Manela, M. Happold and D. Ramanan: Hierarchical Deep Stereo Matching on High- Resolution Images. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
104 CCNet 1.74 % 4.98 % 2.28 % 100.00 % 0.8 s 1 core @ 2.5 Ghz (C/C++)
105 MSFGNet 1.79 % 4.73 % 2.28 % 100.00 % 0.14 s GPU @ >3.5 Ghz (Python)
106 CVANet_RVC 1.76 % 4.91 % 2.28 % 100.00 % 0.8 s 1 core @ 2.5 Ghz (C/C++)
107 PSMNet-Naifan 1.82 % 4.64 % 2.29 % 100.00 % 0.4 s GPU @ 2.5 Ghz (Python)
108 SGNet 1.82 % 4.69 % 2.30 % 100.00 % 0.3 s 1 core @ 2.5 Ghz (Python)
109 DWA code 1.99 % 3.92 % 2.31 % 100.00 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
110 CFP-Net code 1.90 % 4.39 % 2.31 % 100.00 % 0.9 s 8 cores @ 2.5 Ghz (Python)
Z. Zhu, M. He, Y. Dai, Z. Rao and B. Li: Multi-scale Cross-form Pyramid Network for Stereo Matching. arXiv preprint 2019.
111 PSMNet code 1.86 % 4.62 % 2.32 % 100.00 % 0.41 s Nvidia GTX Titan Xp
J. Chang and Y. Chen: Pyramid Stereo Matching Network. arXiv preprint arXiv:1803.08669 2018.
112 GANetREF_RVC code 1.88 % 4.58 % 2.33 % 100.00 % 1.62 s GPU @ >3.5 Ghz (Python + C/C++)
F. Zhang, V. Prisacariu, R. Yang and P. Torr: GA-Net: Guided Aggregation Net for End- to-end Stereo Matching. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2019.
113 DEINet+ 1.82 % 4.88 % 2.33 % 100.00 % 0.21 s GPU @ 2.5 Ghz (Python + C/C++)
114 JPSMNet 1.93 % 4.40 % 2.34 % 100.00 % 0.47 s GPU @ 2.5 Ghz (Python)
115 PSM+LGF55 1.89 % 4.76 % 2.37 % 100.00 % 0.05 s 1 core @ 2.5 Ghz (C/C++)
116 VH-Net 2.11 % 3.71 % 2.38 % 100.00 % 0.4 s GPU @ >3.5 Ghz (Python)
117 PSM+LGF551 1.88 % 4.91 % 2.39 % 100.00 % 0.4 s 1 core @ 2.5 Ghz (Python)
118 MTLnet 2.07 % 4.01 % 2.39 % 100.00 % 0.09 s RTX 2070(pytorch)
119 MABNet_origin 1.89 % 5.02 % 2.41 % 100.00 % 0.38 s Nvidia rtx2080ti (Python)
120 HybridNet code 1.93 % 4.90 % 2.42 % 100.00 % 0.12 s GPU @ 2.5 Ghz (Python)
121 FBNet 1.96 % 4.86 % 2.45 % 100.00 % 0.6 s 8 cores @ 2.5 Ghz (Python)
122 MSN 1.97 % 5.00 % 2.47 % 100.00 % 1.3 s 8 cores @ 2.5 Ghz (Python)
123 ANM3 1.95 % 5.19 % 2.49 % 100.00 % 0.4 s 1 core @ 2.5 Ghz (Python)
124 ANM1 1.99 % 5.05 % 2.50 % 100.00 % 0.41 s 1 core @ 2.5 Ghz (Python)
125 ERSCNet 2.11 % 4.46 % 2.50 % 100.00 % 0.28 s GPU @ 2.5 Ghz (Python)
Anonymous: ERSCNet. Proceedings of the European Conference on Computer Vision (ECCV) 2020.
126 MDA-Net 2.12 % 4.63 % 2.54 % 100.00 % 0.7 s 1 core @ 2.5 Ghz (Python)
127 UberATG-DRISF
This method uses optical flow information.
 2.16 % 4.49 % 2.55 % 100.00 % 0.75 s CPU+GPU @ 2.5 Ghz (Python)
W. Ma, S. Wang, R. Hu, Y. Xiong and R. Urtasun: Deep Rigid Instance Scene Flow. CVPR 2019.
128 AANet code 1.99 % 5.39 % 2.55 % 100.00 % 0.062 s NVIDIA V100 GPU
H. Xu and J. Zhang: AANet: Adaptive Aggregation Network for Efficient Stereo Matching. CVPR 2020.
129 PDSNet 2.29 % 4.05 % 2.58 % 100.00 % 0.5 s 1 core @ 2.5 Ghz (Python)
S. Tulyakov, A. Ivanov and F. Fleuret: Practical Deep Stereo (PDS): Toward applications-friendly deep stereo matching. Proceedings of the international conference on Neural Information Processing Systems (NIPS) 2018.
130 SPP 2.10 % 5.02 % 2.59 % 100.00 % 0.41 s 4 cores @ 2.0 Ghz (Python)
131 DeepPruner (fast) code 2.32 % 3.91 % 2.59 % 100.00 % 0.06 s 1 core @ 2.5 Ghz (C/C++)
S. Duggal, S. Wang, W. Ma, R. Hu and R. Urtasun: DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch. ICCV 2019.
132 SCV code 2.22 % 4.53 % 2.61 % 100.00 % 0.36 s Nvidia GTX 1080 Ti
C. Lu, H. Uchiyama, D. Thomas, A. Shimada and R. Taniguchi: Sparse Cost Volume for Efficient Stereo Matching. Remote Sensing 2018.
133 MS-PSMNet 2.15 % 5.01 % 2.63 % 100.00 % 0.9 s GPU @ 2.5 Ghz (Python)
134 WaveletStereo: 2.24 % 4.62 % 2.63 % 100.00 % 0.27 s 1 core @ 2.5 Ghz (C/C++)
Anonymous: WaveletStereo: Learning wavelet coefficients for stereo matching. arXiv: Computer Vision and Pattern Recognition 2019.
135 AANet_RVC 2.23 % 4.89 % 2.67 % 100.00 % 0.1 s GPU @ 2.5 Ghz (Python)
H. Xu and J. Zhang: AANet: Adaptive Aggregation Network for Efficient Stereo Matching. CVPR 2020.
136 CRL code 2.48 % 3.59 % 2.67 % 100.00 % 0.47 s Nvidia GTX 1080
J. Pang, W. Sun, J. Ren, C. Yang and Q. Yan: Cascade residual learning: A two-stage convolutional neural network for stereo matching. ICCV Workshop on Geometry Meets Deep Learning 2017.
137 GC+CRF 2.11 % 5.71 % 2.71 % 100.00 % 0.27 s GPU @ 2.0 Ghz (C/C++)
138 FADNet code 2.68 % 3.50 % 2.82 % 100.00 % 0.05 s Tesla V100 (Python)
Q. Wang, S. Shi, S. Zheng, K. Zhao and X. Chu: FADNet: A Fast and Accurate Network for Disparity Estimation. arXiv preprint arXiv:2003.10758 2020.
139 PS-Net 2.39 % 4.98 % 2.82 % 100.00 % 0.2 s GPU @ 2.5 Ghz (Python)
140 NVstereo2D 2.57 % 4.20 % 2.84 % 100.00 % 0.01 s GPU @ 2.5 Ghz (Python)
141 GC-NET 2.21 % 6.16 % 2.87 % 100.00 % 0.9 s Nvidia GTX Titan X
A. Kendall, H. Martirosyan, S. Dasgupta, P. Henry, R. Kennedy, A. Bachrach and A. Bry: End-to-End Learning of Geometry and Context for Deep Stereo Regression. Proceedings of the International Conference on Computer Vision (ICCV) 2017.
142 BSDCNet 2.49 % 4.98 % 2.90 % 100.00 % 0.025s 1 core @ 2.5 Ghz (C/C++)
143 MBFnet code 2.59 % 4.80 % 2.96 % 100.00 % 0.05 s GPU @ GTX 2070 (Pytorch)
144 CRAR 2.48 % 5.78 % 3.03 % 100.00 % 0.028 s GTX1080Ti (Pytorch)
145 LRCR 2.55 % 5.42 % 3.03 % 100.00 % 49.2 s Nvidia GTX Titan X
Z. Jie, P. Wang, Y. Ling, B. Zhao, Y. Wei, J. Feng and W. Liu: Left-Right Comparative Recurrent Model for Stereo Matching. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2018.
146 DCNet 2.70 % 4.70 % 3.04 % 100.00 % 0.025s GPU @ Nvidia GTX 1080 (Tensorflow)
147 Stereo Object 1.88 % 8.91 % 3.05 % 100.00 % 0.9 s 8 cores @ 2.5 Ghz (Python)
148 LFENet 2.68 % 4.95 % 3.06 % 100.00 % 0.09 s GPU @ 2.5 Ghz (Python)
ERROR: Wrong syntax in BIBTEX file.
149 Fast DS-CS code 2.83 % 4.31 % 3.08 % 100.00 % 0.02 s GPU @ 2.0 Ghz (Python + C/C++)
K. Yee and A. Chakrabarti: Fast Deep Stereo with 2D Convolutional Processing of Cost Signatures. WACV 2020 (to appear).
150 FDNet 2.83 % 4.31 % 3.08 % 100.00 % 0.7 s 1 core @ 2.5 Ghz (C/C++)
151 RecResNet code 2.46 % 6.30 % 3.10 % 100.00 % 0.3 s GPU @ NVIDIA TITAN X (Tensorflow)
K. Batsos and P. Mordohai: RecResNet: A Recurrent Residual CNN Architecture for Disparity Map Enhancement. In International Conference on 3D Vision (3DV) 2018.
152 NVStereoNet code 2.62 % 5.69 % 3.13 % 100.00 % 0.6 s NVIDIA Titan Xp
N. Smolyanskiy, A. Kamenev and S. Birchfield: On the Importance of Stereo for Accurate Depth Estimation: An Efficient Semi-Supervised Deep Neural Network Approach. arXiv preprint arXiv:1803.09719 2018.
153 NineNet2 2.83 % 4.64 % 3.13 % 100.00 % 0.07 s 1 core @ 2.5 Ghz (C/C++)
154 PMY-net 2.63 % 5.72 % 3.15 % 100.00 % 1 s 1 core @ 2.5 Ghz (Python)
155 LANet 2.69 % 5.43 % 3.15 % 100.00 % 0.08 s GPU @ 2.5 Ghz (Python)
156 Net3_2015 2.69 % 5.44 % 3.15 % 100.00 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
157 AdaStereo 2.64 % 5.75 % 3.16 % 100.00 % 0.41 s GPU @ 2.5 Ghz (Python)
X. Song, G. Yang, X. Zhu, H. Zhou, Z. Wang and J. Shi: AdaStereo: A Simple and Efficient Approach for Adaptive Stereo Matching. arXiv preprint arXiv:2004.04627 2020.
158 DRR 2.58 % 6.04 % 3.16 % 100.00 % 0.4 s Nvidia GTX Titan X
S. Gidaris and N. Komodakis: Detect, Replace, Refine: Deep Structured Prediction For Pixel Wise Labeling. arXiv preprint arXiv:1612.04770 2016.
159 NineNet3 2.69 % 5.57 % 3.17 % 100.00 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
160 Net3_2015 2.71 % 5.55 % 3.18 % 100.00 % 0.3 s 1 core @ 2.5 Ghz (Python + C/C++)
161 MA-Net 2.67 % 5.99 % 3.22 % 100.00 % 0.5 s GPU @ 2.5 Ghz (Python)
162 NineNet 2.70 % 5.94 % 3.24 % 100.00 % 0.07 s 1 core @ 2.5 Ghz (C/C++)
163 DMSNet 2.81 % 5.39 % 3.24 % 100.00 % 0.015625 s 1 core @ 2.5 Ghz (Python)
164 FPN 2.85 % 5.37 % 3.27 % 100.00 % 1 s 1 core @ 2.5 Ghz (Python)
165 SFFNet 2.69 % 6.23 % 3.28 % 100.00 % 0.07 s GPU @ 2.5 Ghz (Python)
166 MS-GCNet 2.58 % 6.83 % 3.29 % 100.00 % 3 s GTX 1080Ti @3.0 Ghz (Python + C/C++)
167 DMSNetv2 2.80 % 5.85 % 3.31 % 100.00 % 0.02 s 1 core @ 2.5 Ghz (Python)
168 DWARF
This method uses optical flow information.
 3.20 % 3.94 % 3.33 % 100.00 % 0.14s - 1.43s TitanXP - JetsonTX2
F. Aleotti, M. Poggi, F. Tosi and S. Mattoccia: Learning end-to-end scene flow by distilling single tasks knowledge. Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20) 2020.
169 AbNet1 3.14 % 4.43 % 3.35 % 100.00 % 0.02 s 1 core @ 2.5 Ghz (C/C++)
170 SsSMnet 2.70 % 6.92 % 3.40 % 100.00 % 0.8 s P100
Y. Zhong, Y. Dai and H. Li: Self-Supervised Learning for Stereo Matching with Self-Improving Ability. arXiv:1709.00930 2017.
171 L-ResMatch code 2.72 % 6.95 % 3.42 % 100.00 % 48 s 1 core @ 2.5 Ghz (C/C++)
A. Shaked and L. Wolf: Improved Stereo Matching with Constant Highway Networks and Reflective Loss. arXiv preprint arxiv:1701.00165 2016.
172 Displets v2 code 3.00 % 5.56 % 3.43 % 100.00 % 265 s >8 cores @ 3.0 Ghz (Matlab + C/C++)
F. Guney and A. Geiger: Displets: Resolving Stereo Ambiguities using Object Knowledge. Conference on Computer Vision and Pattern Recognition (CVPR) 2015.
173 LBPS code 2.85 % 6.35 % 3.44 % 100.00 % 0.39 s GPU @ 2.5 Ghz (C/C++)
P. Knöbelreiter, C. Sormann, A. Shekhovtsov, F. Fraundorfer and T. Pock: Belief Propagation Reloaded: Learning BP-Layers for Labeling Problems. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020.
174 TBDN_Net 2.86 % 6.92 % 3.53 % 100.00 % 1.0 s GPU @ 2.5 Ghz (Python)
175 ACOSF
This method uses optical flow information.
 2.79 % 7.56 % 3.58 % 100.00 % 5 min 1 core @ 3.0 Ghz (Matlab + C/C++)
C. Li, H. Ma and Q. Liao: Two-Stage Adaptive Object Scene Flow Using Hybrid CNN-CRF Model. International Conference on Pattern Recognition (ICPR) 2020.
176 ZYY 2.96 % 6.74 % 3.58 % 100.00 % 0.22 s 1 core @ 2.5 Ghz (C/C++)
177 TBDN_Net40 2.89 % 7.15 % 3.60 % 100.00 % 0.95 s GPU @ 2.5 Ghz (Python)
178 CNNF+SGM 2.78 % 7.69 % 3.60 % 100.00 % 71 s TESLA K40C
F. Zhang and B. Wah: Fundamental Principles on Learning New Features for Effective Dense Matching. IEEE Transactions on Image Processing 2018.
179 PBCP 2.58 % 8.74 % 3.61 % 100.00 % 68 s Nvidia GTX Titan X
A. Seki and M. Pollefeys: Patch Based Confidence Prediction for Dense Disparity Map. British Machine Vision Conference (BMVC) 2016.
180 ASMNet 3.18 % 5.98 % 3.64 % 100.00 % 0.04 s 4 cores @ 2.5 Ghz (Python)
181 SGM-Net 2.66 % 8.64 % 3.66 % 100.00 % 67 s Titan X
A. Seki and M. Pollefeys: SGM-Nets: Semi-Global Matching With Neural Networks. CVPR 2017.
182 DSMNet 3.11 % 6.72 % 3.71 % 100.00 % 1.5 s GPU @ 2.5 Ghz (Python)
183 Three3 2.99 % 7.33 % 3.71 % 100.00 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
184 HSM-Net_RVC code 2.74 % 8.73 % 3.74 % 100.00 % 0.97 s GPU @ 2.5 Ghz (Python)
G. Yang, J. Manela, M. Happold and D. Ramanan: Hierarchical deep stereo matching on high-resolution images. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2019.
185 MABNet_tiny 3.04 % 8.07 % 3.88 % 100.00 % 0.11 s Nvidia rtx2080ti (Python)
186 MC-CNN-acrt code 2.89 % 8.88 % 3.89 % 100.00 % 67 s Nvidia GTX Titan X (CUDA, Lua/Torch7)
J. Zbontar and Y. LeCun: Stereo Matching by Training a Convolutional Neural Network to Compare Image Patches. Submitted to JMLR .
187 Net3_3 3.52 % 6.22 % 3.97 % 100.00 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
188 Reversing-PSMNet code 3.13 % 8.70 % 4.06 % 100.00 % 0.41 s 1 core @ 1.5 Ghz (Python)
F. Aleotti, F. Tosi, L. Zhang, M. Poggi and S. Mattoccia: Reversing the cycle: self-supervised deep stereo through enhanced monocular distillation. European Conference on Computer Vision (ECCV) 2020.
189 DAStereo 3.67 % 6.83 % 4.20 % 100.00 % 0.32 s 1 core @ 2.5 Ghz (Python)
190 STTRV1_RVC 3.56 % 7.71 % 4.25 % 99.99 % 0.6 s GPU @ 1.0 Ghz (Python)
191 PRSM
This method uses optical flow information.
This method makes use of multiple (>2) views.
 code 3.02 % 10.52 % 4.27 % 99.99 % 300 s 1 core @ 2.5 Ghz (C/C++)
C. Vogel, K. Schindler and S. Roth: 3D Scene Flow Estimation with a Piecewise Rigid Scene Model. ijcv 2015.
192 DispNetC code 4.32 % 4.41 % 4.34 % 100.00 % 0.06 s Nvidia GTX Titan X (Caffe)
N. Mayer, E. Ilg, P. Häusser, P. Fischer, D. Cremers, A. Dosovitskiy and T. Brox: A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation. CVPR 2016.
193 SGM-Forest 3.11 % 10.74 % 4.38 % 99.92 % 6 seconds 1 core @ 3.0 Ghz (Python/C/C++)
J. Schönberger, S. Sinha and M. Pollefeys: Learning to Fuse Proposals from Multiple Scanline Optimizations in Semi-Global Matching. European Conference on Computer Vision (ECCV) 2018.
194 SSF
This method uses optical flow information.
 3.55 % 8.75 % 4.42 % 100.00 % 5 min 1 core @ 2.5 Ghz (Matlab + C/C++)
Z. Ren, D. Sun, J. Kautz and E. Sudderth: Cascaded Scene Flow Prediction using Semantic Segmentation. International Conference on 3D Vision (3DV) 2017.
195 ISF
This method uses optical flow information.
 4.12 % 6.17 % 4.46 % 100.00 % 10 min 1 core @ 3 Ghz (C/C++)
A. Behl, O. Jafari, S. Mustikovela, H. Alhaija, C. Rother and A. Geiger: Bounding Boxes, Segmentations and Object Coordinates: How Important is Recognition for 3D Scene Flow Estimation in Autonomous Driving Scenarios?. International Conference on Computer Vision (ICCV) 2017.
196 USegScene 4.12 % 6.58 % 4.53 % 100.00 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
197 Content-CNN 3.73 % 8.58 % 4.54 % 100.00 % 1 s Nvidia GTX Titan X (Torch)
W. Luo, A. Schwing and R. Urtasun: Efficient Deep Learning for Stereo Matching. CVPR 2016.
198 MADnet code 3.75 % 9.20 % 4.66 % 100.00 % 0.02 s GPU @ 2.5 Ghz (Python)
A. Tonioni, F. Tosi, M. Poggi, S. Mattoccia and L. Di Stefano: Real-Time self-adaptive deep stereo. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
199 DTF
This method uses optical flow information.
This method makes use of multiple (>2) views.
 3.91 % 8.57 % 4.68 % 100.00 % 0.14 s RTX 2080 Ti
200 VN 4.29 % 7.65 % 4.85 % 100.00 % 0.5 s GPU @ 3.5 Ghz (Python + C/C++)
P. Knöbelreiter and T. Pock: Learned Collaborative Stereo Refinement. German Conference on Pattern Recognition (GCPR) 2019.
201 LWANet 4.28 % 8.22 % 4.94 % 100.00 % 0.02 s GPU @ 2.5 Ghz (Python)
202 naive-stereo 4.43 % 7.65 % 4.96 % 100.00 % 0.05 s 1 core @ 2.5 Ghz (Python)
203 MC-CNN-WS code 3.78 % 10.93 % 4.97 % 100.00 % 1.35 s 1 core 2.5 Ghz + K40 NVIDIA, Lua-Torch
S. Tulyakov, A. Ivanov and F. Fleuret: Weakly supervised learning of deep metrics for stereo reconstruction. ICCV 2017.
204 3DMST 3.36 % 13.03 % 4.97 % 100.00 % 93 s 1 core @ >3.5 Ghz (C/C++)
X. Lincheng Li and L. Zhang: 3D Cost Aggregation with Multiple Minimum Spanning Trees for Stereo Matching. submitted to Applied Optics .
205 CBMV_ROB code 3.55 % 12.09 % 4.97 % 100.00 % 250 s 6 core @ 3.0 Ghz (Python + C/C++)
K. Batsos, C. Cai and P. Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2018.
206 SMV 3.80 % 10.99 % 5.00 % 100.00 % 0.3 s 6 core @ 3.6 Ghz (Python)
207 OSF+TC
This method uses optical flow information.
This method makes use of multiple (>2) views.
 4.11 % 9.64 % 5.03 % 100.00 % 50 min 1 core @ 2.5 Ghz (C/C++)
M. Neoral and J. Šochman: Object Scene Flow with Temporal Consistency. 22nd Computer Vision Winter Workshop (CVWW) 2017.
208 CBMV code 4.17 % 9.53 % 5.06 % 100.00 % 250 s 6 cores @ 3.0 Ghz (Python,C/C++,CUDA Nvidia TitanX)
K. Batsos, C. Cai and P. Mordohai: CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation. 2018.
209 SimpleStereo 4.48 % 8.29 % 5.12 % 100.00 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
210 PWOC-3D
This method uses optical flow information.
 code 4.19 % 9.82 % 5.13 % 100.00 % 0.13 s GTX 1080 Ti
R. Saxena, R. Schuster, O. Wasenmüller and D. Stricker: PWOC-3D: Deep Occlusion-Aware End-to-End Scene Flow Estimation. Intelligent Vehicles Symposium (IV) 2019.
211 naive-stereo-v1 4.99 % 6.56 % 5.25 % 100.00 % 0.05 s 1 core @ 2.5 Ghz (Python)
212 OSF 2018
This method uses optical flow information.
 code 4.11 % 11.12 % 5.28 % 100.00 % 390 s 1 core @ 2.5 Ghz (Matlab + C/C++)
M. Menze, C. Heipke and A. Geiger: Object Scene Flow. ISPRS Journal of Photogrammetry and Remote Sensing (JPRS) 2018.
213 SPS-St code 3.84 % 12.67 % 5.31 % 100.00 % 2 s 1 core @ 3.5 Ghz (C/C++)
K. Yamaguchi, D. McAllester and R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014.
214 LGF 4.78 % 8.10 % 5.33 % 100.00 % 0.06 s GPU @ 2.5 Ghz (Python)
215 MDP
This method uses stereo information.
 4.19 % 11.25 % 5.36 % 100.00 % 11.4 s 4 cores @ 3.5 Ghz (Matlab + C/C++)
A. Li, D. Chen, Y. Liu and Z. Yuan: Coordinating Multiple Disparity Proposals for Stereo Computation. IEEE Conference on Computer Vision and Pattern Recognition 2016.
216 LGF_dense 4.68 % 9.14 % 5.42 % 100.00 % 0.06 s 1 core @ 2.5 Ghz (C/C++)
217 SFF++
This method uses optical flow information.
This method makes use of multiple (>2) views.
 4.27 % 12.38 % 5.62 % 100.00 % 78 s 4 cores @ 3.5 Ghz (C/C++)
R. Schuster, O. Wasenmüller, C. Unger, G. Kuschk and D. Stricker: SceneFlowFields++: Multi-frame Matching, Visibility Prediction, and Robust Interpolation for Scene Flow Estimation. International Journal of Computer Vision (IJCV) 2019.
218 OSF
This method uses optical flow information.
 code 4.54 % 12.03 % 5.79 % 100.00 % 50 min 1 core @ 2.5 Ghz (C/C++)
M. Menze and A. Geiger: Object Scene Flow for Autonomous Vehicles. Conference on Computer Vision and Pattern Recognition (CVPR) 2015.
219 pSGM 4.84 % 11.64 % 5.97 % 100.00 % 7.77 s 4 cores @ 3.5 Ghz (C/C++)
Y. Lee, M. Park, Y. Hwang, Y. Shin and C. Kyung: Memory-Efficient Parametric Semiglobal Matching. IEEE Signal Processing Letters 2018.
220 CSF
This method uses optical flow information.
 4.57 % 13.04 % 5.98 % 99.99 % 80 s 1 core @ 2.5 Ghz (C/C++)
Z. Lv, C. Beall, P. Alcantarilla, F. Li, Z. Kira and F. Dellaert: A Continuous Optimization Approach for Efficient and Accurate Scene Flow. European Conf. on Computer Vision (ECCV) 2016.
221 MBM 4.69 % 13.05 % 6.08 % 100.00 % 0.13 s 1 core @ 3.0 Ghz (C/C++)
N. Einecke and J. Eggert: A Multi-Block-Matching Approach for Stereo. IV 2015.
222 PR-Sceneflow
This method uses optical flow information.
 code 4.74 % 13.74 % 6.24 % 100.00 % 150 s 4 core @ 3.0 Ghz (Matlab + C/C++)
C. Vogel, K. Schindler and S. Roth: Piecewise Rigid Scene Flow. ICCV 2013.
223 DispSegNet 4.20 % 16.97 % 6.33 % 100.00 % 0.9 s GPU @ 2.5 Ghz (Python)
J. Zhang, K. Skinner, R. Vasudevan and M. Johnson-Roberson: DispSegNet: Leveraging Semantics for End- to-End Learning of Disparity Estimation From Stereo Imagery. IEEE Robotics and Automation Letters 2019.
224 DeepCostAggr code 5.34 % 11.35 % 6.34 % 99.98 % 0.03 s GPU @ 2.5 Ghz (C/C++)
A. Kuzmin, D. Mikushin and V. Lempitsky: End-to-end Learning of Cost-Volume Aggregation for Real-time Dense Stereo. 2017 IEEE 27th International Workshop on Machine Learning for Signal Processing (MLSP) 2017.
225 SGM_RVC 5.06 % 13.00 % 6.38 % 100.00 % 0.11 s Nvidia GTX 980
H. Hirschm\"uller: Stereo Processing by Semi-Global Matching and Mutual Information. IEEE Transactions on Pattern Analysis and Machine Intelligence 2008.
226 OASM-DDS 4.66 % 15.76 % 6.51 % 100.00 % 0.90 s 1 core @ 2.5 Ghz (Python)
227 SceneFFields
This method uses optical flow information.
 5.12 % 13.83 % 6.57 % 100.00 % 65 s 4 cores @ 3.7 Ghz (C/C++)
R. Schuster, O. Wasenmüller, G. Kuschk, C. Bailer and D. Stricker: SceneFlowFields: Dense Interpolation of Sparse Scene Flow Correspondences. IEEE Winter Conference on Applications of Computer Vision (WACV) 2018.
228 SPS+FF++
This method uses optical flow information.
 code 5.47 % 12.19 % 6.59 % 100.00 % 36 s 1 core @ 3.5 Ghz (C/C++)
R. Schuster, O. Wasenmüller and D. Stricker: Dense Scene Flow from Stereo Disparity and Optical Flow. ACM Computer Science in Cars Symposium (CSCS) 2018.
229 Flow2Stereo 5.01 % 14.62 % 6.61 % 99.97 % 0.05 s GPU @ 2.5 Ghz (Python)
P. Liu, I. King, M. Lyu and J. Xu: Flow2Stereo: Effective Self-Supervised Learning of Optical Flow and Stereo Matching. CVPR 2020.
230 FSF+MS
This method uses optical flow information.
This method makes use of the epipolar geometry.
This method makes use of multiple (>2) views.
 5.72 % 11.84 % 6.74 % 100.00 % 2.7 s 4 cores @ 3.5 Ghz (C/C++)
T. Taniai, S. Sinha and Y. Sato: Fast Multi-frame Stereo Scene Flow with Motion Segmentation. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2017) 2017.
231 AABM 4.88 % 16.07 % 6.74 % 100.00 % 0.08 s 1 core @ 3.0 Ghz (C/C++)
N. Einecke and J. Eggert: Stereo Image Warping for Improved Depth Estimation of Road Surfaces. IV 2013.
232 DistillFlow
This method uses optical flow information.
 5.00 % 15.88 % 6.81 % 100.00 % 0.03 s 1 core @ 2.5 Ghz (C/C++)
233 SGM+C+NL
This method uses optical flow information.
 code 5.15 % 15.29 % 6.84 % 100.00 % 4.5 min 1 core @ 2.5 Ghz (C/C++)
H. Hirschmüller: Stereo Processing by Semiglobal Matching and Mutual Information. PAMI 2008.
D. Sun, S. Roth and M. Black: A Quantitative Analysis of Current Practices in Optical Flow Estimation and the Principles Behind Them. IJCV 2013.
234 SGM+LDOF
This method uses optical flow information.
 code 5.15 % 15.29 % 6.84 % 100.00 % 86 s 1 core @ 2.5 Ghz (C/C++)
H. Hirschmüller: Stereo Processing by Semiglobal Matching and Mutual Information. PAMI 2008.
T. Brox and J. Malik: Large Displacement Optical Flow: Descriptor Matching in Variational Motion Estimation. PAMI 2011.
235 SGM+SF
This method uses optical flow information.
 5.15 % 15.29 % 6.84 % 100.00 % 45 min 16 core @ 3.2 Ghz (C/C++)
H. Hirschmüller: Stereo Processing by Semiglobal Matching and Mutual Information. PAMI 2008.
M. Hornacek, A. Fitzgibbon and C. Rother: SphereFlow: 6 DoF Scene Flow from RGB-D Pairs. CVPR 2014.
236 SNCC 5.36 % 16.05 % 7.14 % 100.00 % 0.08 s 1 core @ 3.0 Ghz (C/C++)
N. Einecke and J. Eggert: A Two-Stage Correlation Method for Stereoscopic Depth Estimation. DICTA 2010.
237 MBSF
This method uses optical flow information.
 5.63 % 15.04 % 7.20 % 100.00 % 1 min GPU @ 2.5 Ghz (C/C++)
238 PASMnet code 5.41 % 16.36 % 7.23 % 100.00 % 0.5 s GPU @ 2.5 Ghz (Python)
L. Wang, Y. Guo, Y. Wang, Z. Liang, Z. Lin, J. Yang and W. An: Parallax Attention for Unsupervised Stereo Correspondence Learning. IEEE Transactions on Pattern Analysis and Machine Intelligence(T-PAMI) 2020.
239 AAFS 6.27 % 13.95 % 7.54 % 100.00 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
240 NL-DL 6.73 % 13.08 % 7.79 % 100.00 % 0.04 s 8 cores @ 2.5 Ghz (C/C++)
241 GOUSM 6.42 % 17.24 % 8.22 % 100.00 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
242 CSCT+SGM+MF 6.91 % 14.87 % 8.24 % 100.00 % 0.0064 s Nvidia GTX Titan X @ 1.0 Ghz (CUDA)
D. Hernandez-Juarez, A. Chacon, A. Espinosa, D. Vazquez, J. Moure and A. Lopez: Embedded real-time stereo estimation via Semi-Global Matching on the GPU. Procedia Computer Science 2016.
243 MeshStereo code 5.82 % 21.21 % 8.38 % 100.00 % 87 s 1 core @ 2.5 Ghz (C/C++)
C. Zhang, Z. Li, Y. Cheng, R. Cai, H. Chao and Y. Rui: MeshStereo: A Global Stereo Model With Mesh Alignment Regularization for View Interpolation. The IEEE International Conference on Computer Vision (ICCV) 2015.
244 PCOF + ACTF
This method uses optical flow information.
 6.31 % 19.24 % 8.46 % 100.00 % 0.08 s GPU @ 2.0 Ghz (C/C++)
M. Derome, A. Plyer, M. Sanfourche and G. Le Besnerais: A Prediction-Correction Approach for Real-Time Optical Flow Computation Using Stereo. German Conference on Pattern Recognition 2016.
245 PCOF-LDOF
This method uses optical flow information.
 6.31 % 19.24 % 8.46 % 100.00 % 50 s 1 core @ 3.0 Ghz (C/C++)
M. Derome, A. Plyer, M. Sanfourche and G. Le Besnerais: A Prediction-Correction Approach for Real-Time Optical Flow Computation Using Stereo. German Conference on Pattern Recognition 2016.
246 OASM-Net 6.89 % 19.42 % 8.98 % 100.00 % 0.73 s GPU @ 2.5 Ghz (Python)
A. Li and Z. Yuan: Occlusion Aware Stereo Matching via Cooperative Unsupervised Learning. Proceedings of the Asian Conference on Computer Vision, ACCV 2018.
247 ELAS_RVC code 7.38 % 21.15 % 9.67 % 100.00 % 0.19 s 4 cores @ >3.5 Ghz (C/C++)
A. Geiger, M. Roser and R. Urtasun: Efficient Large-Scale Stereo Matching. ACCV 2010.
248 DDP_out_ELAS_params1 7.45 % 21.11 % 9.72 % 100.00 % 30 min GPU @ 2.5 Ghz (Python)
249 ELAS code 7.86 % 19.04 % 9.72 % 92.35 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
A. Geiger, M. Roser and R. Urtasun: Efficient Large-Scale Stereo Matching. ACCV 2010.
250 DDP_in_ELAS_params1 7.48 % 21.14 % 9.76 % 99.97 % 1 s 1 core @ 2.5 Ghz (Matlab + C/C++)
251 REAF code 8.43 % 18.51 % 10.11 % 100.00 % 1.1 s 1 core @ 2.5 Ghz (C/C++)
C. Cigla: Recursive Edge-Aware Filters for Stereo Matching. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops 2015.
252 iGF
This method makes use of multiple (>2) views.
 8.64 % 21.85 % 10.84 % 100.00 % 220 s 1 core @ 3.0 Ghz (C/C++)
R. Hamzah, H. Ibrahim and A. Hassan: Stereo matching algorithm based on per pixel difference adjustment, iterative guided filter and graph segmentation. Journal of Visual Communication and Image Representation 2016.
253 OCV-SGBM code 8.92 % 20.59 % 10.86 % 90.41 % 1.1 s 1 core @ 2.5 Ghz (C/C++)
H. Hirschmueller: Stereo processing by semiglobal matching and mutual information. PAMI 2008.
254 PPEP-GF code 9.87 % 19.01 % 11.39 % 100.00 % 3.41 s 2 cores @ 3.0 Ghz (C/C++)
ERROR: Wrong syntax in BIBTEX file.
255 TW-SMNet 11.92 % 12.16 % 11.96 % 100.00 % 0.7 s GPU @ 2.5 Ghz (Python)
M. El-Khamy, H. Ren, X. Du and J. Lee: TW-SMNet: Deep Multitask Learning of Tele-Wide Stereo Matching. arXiv:1906.04463 2019.
256 SDM 9.41 % 24.75 % 11.96 % 62.56 % 1 min 1 core @ 2.5 Ghz (C/C++)
J. Kostkova: Stratified dense matching for stereopsis in complex scenes. BMVC 2003.
257 SGM&FlowFie+
This method uses optical flow information.
 11.93 % 20.57 % 13.37 % 81.24 % 29 s 1 core @ 3.5 Ghz (C/C++)
R. Schuster, C. Bailer, O. Wasenmüller and D. Stricker: Combining Stereo Disparity and Optical Flow for Basic Scene Flow. Commercial Vehicle Technology Symposium (CVTS) 2018.
258 DDP 12.32 % 20.20 % 13.63 % 100.00 % 10 min 1 core @ 2.5 Ghz (Python)
259 GCSF
This method uses optical flow information.
 code 11.64 % 27.11 % 14.21 % 100.00 % 2.4 s 1 core @ 2.5 Ghz (C/C++)
J. Cech, J. Sanchez-Riera and R. Horaud: Scene Flow Estimation by growing Correspondence Seeds. CVPR 2011.
260 MT-TW-SMNet 15.47 % 16.25 % 15.60 % 100.00 % 0.4s GPU @ 2.5 Ghz (Python)
M. El-Khamy, X. Du, H. Ren and J. Lee: Multi-Task Learning of Depth from Tele and Wide Stereo Image Pairs. Proceedings of the IEEE Conference on Image Processing 2019.
261 Mono-SF
This method uses optical flow information.
 14.21 % 26.94 % 16.32 % 100.00 % 41 s 1 core @ 3.5 Ghz (Matlab + C/C++)
F. Brickwedde, S. Abraham and R. Mester: Mono-SF: Multi-View Geometry meets Single-View Depth for Monocular Scene Flow Estimation of Dynamic Traffic Scenes. Proc. of International Conference on Computer Vision (ICCV) 2019.
262 AbNet2 14.71 % 29.88 % 17.23 % 100.00 % 0.01 s 1 core @ 2.5 Ghz (C/C++)
263 DDP_out_ELAS_params2 16.28 % 26.06 % 17.91 % 100.00 % 10 min 1 core @ 2.5 Ghz (Python)
264 CostFilter code 17.53 % 22.88 % 18.42 % 100.00 % 4 min 1 core @ 2.5 Ghz (Matlab)
C. Rhemann, A. Hosni, M. Bleyer, C. Rother and M. Gelautz: Fast Cost-Volume Filtering for Visual Correspondence and Beyond. CVPR 2011.
265 MonoComb
This method uses optical flow information.
 17.89 % 21.16 % 18.44 % 100.00 % 0.58 s RTX 2080 Ti
266 DWBSF
This method uses optical flow information.
 19.61 % 22.69 % 20.12 % 100.00 % 7 min 4 cores @ 3.5 Ghz (C/C++)
C. Richardt, H. Kim, L. Valgaerts and C. Theobalt: Dense Wide-Baseline Scene Flow From Two Handheld Video Cameras. 3DV 2016.
267 monoResMatch code 22.10 % 19.81 % 21.72 % 100.00 % 0.16 s Titan X GPU
F. Tosi, F. Aleotti, M. Poggi and S. Mattoccia: Learning monocular depth estimation infusing traditional stereo knowledge. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019.
268 DDP_in_ELAS_params2 20.63 % 28.68 % 21.97 % 85.92 % 1 s 1 core @ 2.5 Ghz (Matlab + C/C++)
269 Self-Mono-SF-ft
This method uses optical flow information.
 code 20.72 % 29.41 % 22.16 % 100.00 % 0.09 s NVIDIA GTX 1080 Ti
J. Hur and S. Roth: Self-Supervised Monocular Scene Flow Estimation. CVPR 2020.
270 OCV-BM code 24.29 % 30.13 % 25.27 % 58.54 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
G. Bradski: The OpenCV Library. Dr. Dobb's Journal of Software Tools 2000.
271 Mono expansion
This method uses optical flow information.
 24.85 % 27.90 % 25.36 % 100.00 % 0.25 s GPU @ 2.5 Ghz (Python)
272 VSF
This method uses optical flow information.
 code 27.31 % 21.72 % 26.38 % 100.00 % 125 min 1 core @ 2.5 Ghz (C/C++)
F. Huguet and F. Devernay: A Variational Method for Scene Flow Estimation from Stereo Sequences. ICCV 2007.
273 YunYang-Monodepth 26.06 % 28.92 % 26.53 % 100.00 % 0.01 s 1 core @ 2.5 Ghz (Python)
274 SED code 25.01 % 40.43 % 27.58 % 4.02 % 0.68 s 1 core @ 2.0 Ghz (C/C++)
D. Pe\~{n}a and A. Sutherland: Disparity Estimation by Simultaneous Edge Drawing. Computer Vision -- ACCV 2016 Workshops: ACCV 2016 International Workshops, Taipei, Taiwan, November 20-24, 2016, Revised Selected Papers, Part II 2017.
275 MTS code 26.68 % 47.30 % 30.11 % 4.31 % 1.7 s 4 cores @ 3.5 Ghz (C/C++)
276 mts1 code 28.03 % 46.55 % 31.11 % 2.52 % 0.18 s 4 cores @ 3.5 Ghz (C/C++)
R. Brandt, N. Strisciuglio, N. Petkov and M. Wilkinson: Efficient binocular stereo correspondence matching with 1-D Max-Trees. Pattern Recognition Letters 2020.
277 Self-Mono-SF
This method uses optical flow information.
 code 31.22 % 48.04 % 34.02 % 100.00 % 0.09 s NVIDIA GTX 1080 Ti
J. Hur and S. Roth: Self-Supervised Monocular Scene Flow Estimation. CVPR 2020.
278 MST code 45.83 % 38.22 % 44.57 % 100.00 % 7 s 1 core @ 2.5 Ghz (Matlab + C/C++)
Q. Yang: A Non-Local Cost Aggregation Method for Stereo Matching. CVPR 2012.
Table as LaTeX | Only published Methods




Related Datasets

  • HCI/Bosch Robust Vision Challenge: Optical flow and stereo vision challenge on high resolution imagery recorded at a high frame rate under diverse weather conditions (e.g., sunny, cloudy, rainy). The Robert Bosch AG provides a prize for the best performing method.
  • Image Sequence Analysis Test Site (EISATS): Synthetic image sequences with ground truth information provided by UoA and Daimler AG. Some of the images come with 3D range sensor information.
  • Middlebury Stereo Evaluation: The classic stereo evaluation benchmark, featuring four test images in version 2 of the benchmark, with very accurate ground truth from a structured light system. 38 image pairs are provided in total.
  • Daimler Stereo Dataset: Stereo bad weather highway scenes with partial ground truth for freespace
  • Make3D Range Image Data: Images with small-resolution ground truth used to learn and evaluate depth from single monocular images.
  • Lubor Ladicky's Stereo Dataset: Stereo Images with manually labeled ground truth based on polygonal areas.
  • Middlebury Optical Flow Evaluation: The classic optical flow evaluation benchmark, featuring eight test images, with very accurate ground truth from a shape from UV light pattern system. 24 image pairs are provided in total.

Citation

When using this dataset in your research, we will be happy if you cite us:
@ARTICLE{Menze2018JPRS,
  author = {Moritz Menze and Christian Heipke and Andreas Geiger},
  title = {Object Scene Flow},
  journal = {ISPRS Journal of Photogrammetry and Remote Sensing (JPRS)},
  year = {2018}
}
@INPROCEEDINGS{Menze2015ISA,
  author = {Moritz Menze and Christian Heipke and Andreas Geiger},
  title = {Joint 3D Estimation of Vehicles and Scene Flow},
  booktitle = {ISPRS Workshop on Image Sequence Analysis (ISA)},
  year = {2015}
}



eXTReMe Tracker