Andreas Geiger (KIT) | Philip Lenz (KIT) | Christoph Stiller (KIT) | Raquel Urtasun (TTI-C)

Dataset


The stereo / flow benchmark consists of 194 training image pairs and 195 test image pairs, saved in loss less png format. Our evaluation server computes the average number of bad pixels for all non-occluded or occluded (=all groundtruth) pixels. 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.

Evaluation

Our evaluation table ranks all methods according to the number of non-occluded erroneous pixels at the specified disparity / end-point error threshold. 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. For each method we show:

  • Out-Noc: Percentage of erroneous pixels in non-occluded areas
  • Out-All: Percentage of erroneous pixels in total
  • Avg-Noc: Average disparity / end-point error in non-occluded areas
  • Avg-All: Average disparity / end-point error in total
  • Density: Percentage of pixels for which ground truth has been provided by the method

Note: Our main ranking is computed at 3 pixels error threshold, evaluating all pixels. For methods which do not provide dense result we use background interpolation to fill in missing values. When publishing, please make sure that you provide this 'default' table in your publication.

Error threshold        Evaluation area

Stereo Evaluation


Rank Method Setting Out-Noc Out-All Avg-Noc Avg-All Density Runtime Environment
1 PCBP-SS 3.49 % 4.79 % 0.8 px 1.0 px 100.00 % 5 min 4 cores @ 2.5 Ghz (Matlab + C/C++)
K. Yamaguchi, D. McAllester and R. Urtasun: Robust Monocular Epipolar Flow Estimation. CVPR 2013.
2 StereoSLIC 3.99 % 5.17 % 0.9 px 1.0 px 99.89 % 2.3 s 1 core @ 3.0 Ghz (C/C++)
K. Yamaguchi, D. McAllester and R. Urtasun: Robust Monocular Epipolar Flow Estimation. CVPR 2013.
3 PR-Sf+E 4.09 % 4.95 % 0.9 px 1.0 px 100.00 % 200 s 4 cores @ 3.0 Ghz (Matlab + C/C++)
Anonymous submission
4 PCBP 4.13 % 5.45 % 0.9 px 1.2 px 100.00 % 5 min 4 cores @ 2.5 Ghz (Matlab + C/C++)
K. Yamaguchi, T. Hazan, D. McAllester and R. Urtasun: Continuous Markov Random Fields for Robust Stereo Estimation. ECCV 2012.
5 PR-Sceneflow 4.46 % 5.32 % 1.0 px 1.1 px 100.00 % 150 sec 4 core @ 3.0 Ghz (Matlab - C/C++)
Anonymous submission
6 wSGM 5.03 % 6.24 % 1.3 px 1.6 px 97.03 % 6s 1 core @ 3.5 Ghz (C/C++)
Anonymous submission
7 ATGV 5.05 % 6.91 % 1.0 px 1.6 px 100.00 % 6 min >8 cores @ 3.0 Ghz (Matlab + C/C++)
R. Ranftl, T. Pock and H. Bischof: Minimizing TGV-based Variational Models with Non-Convex Data terms. ICSSVM 2013.
8 iSGM 5.16 % 7.19 % 1.2 px 2.1 px 94.70 % 8 s 2 cores @ 2.5 Ghz (C/C++)
S. Hermann and R. Klette: Iterative Semi-Global Matching for Robust Driver Assistance Systems. ACCV 2012.
9 AABM 5.50 % 6.60 % 1.1 px 1.3 px 100.00 % 0.43 s 1 core @ 3.0 Ghz (C/C++)
N. Einecke and J. Eggert: Appearance-Aligned Block-Matching Stereo. IV 2013.
10 SGM 5.83 % 7.08 % 1.2 px 1.3 px 85.80 % 3.7 s 1 core @ 3.0 Ghz (C/C++)
H. Hirschmueller: Stereo Processing by Semi-Global Matching and Mutual Information. PAMI 2008.
11 TGV2ADC 6.02 % 6.94 % 1.1 px 1.2 px 99.99 % 8 s GPU @ 2.5 Ghz (C/C++)
Anonymous submission
12 CD 6.17 % 7.49 % 1.2 px 1.4 px 100.00 % 5 s 1 core @ 2.5 Ghz (C/C++)
Anonymous submission
13 SNCC 6.27 % 7.33 % 1.4 px 1.5 px 100.00 % 0.27 s 1 core @ 3.0 Ghz (C/C++)
N. Einecke and J. Eggert: A Two-Stage Correlation Method for Stereoscopic Depth Estimation. DICTA 2010.
14 ITGV 6.31 % 7.40 % 1.3 px 1.5 px 100.00 % 7 s 1 core @ 3.0 Ghz (Matlab + C/C++)
R. Ranftl, S. Gehrig, T. Pock and H. Bischof: Pushing the Limits of Stereo Using Variational Stereo Estimation. IV 2012.
15 RWR 6.36 % 7.51 % 1.2 px 1.4 px 100.00 % 1 min 1 core @ 2.5 Ghz (C/C++)
Anonymous submission
16 LDE 6.81 % 8.92 % 1.8 px 2.5 px 100.00 % 14 s 2 cores @ 2.5 Ghz (C/C++)
Anonymous submission
17 BSSM 7.50 % 8.89 % 1.4 px 1.6 px 94.87 % 20.7 s 1 core @ 3.5 Ghz (C/C++)
Anonymous submission
18 OCV-SGBM 7.64 % 9.13 % 1.8 px 2.0 px 86.50 % 1.1 s 1 core @ 2.5 Ghz (C/C++)
H. Hirschmueller: Stereo processing by semiglobal matching and mutual information. PAMI 2008.
19 ELAS 8.24 % 9.95 % 1.4 px 1.6 px 94.55 % 0.3 s 1 core @ 2.5 Ghz (C/C++)
A. Geiger, M. Roser and R. Urtasun: Efficient Large-Scale Stereo Matching. ACCV 2010.
20 linBP 8.66 % 10.81 % 1.7 px 2.7 px 99.89 % 1.6 min 1 core @ 3.0 Ghz (C/C++)
W. Khan, V. Suaste, D. Caudillo and R. Klette: Belief Propagation Stereo Matching Compared to iSGM on Binocular or Trinocular Video Data. IV 2013.
21 SM_GPTM 9.86 % 11.45 % 2.1 px 2.6 px 100.00 % 6.5 s 2 cores @ 2.5 Ghz (C/C++)
C. Cigla and A. Alatan: An Improved Stereo Matching Algorithm with Ground Plane and Temporal Smoothness Constraints. ECCV Workshops 2012.
22 LAMC-DSΜ 9.90 % 11.56 % 2.1 px 2.7 px 99.96 % 10.8 min 2 cores @ 2.5 Ghz (Matlab)
C. Stentoumis, L. Grammatikopoulos, I. Kalisperakis, E. Petsa and G. Karras: A local adaptive approach for dense stereo matching in architectural scene reconstruction. ISPRS 2013.
23 MS-DSI 10.68 % 12.11 % 1.9 px 2.2 px 100.00 % 10.3 s >8 cores @ 2.5 Ghz (C/C++)
Anonymous submission
24 SDM 10.98 % 12.19 % 2.0 px 2.3 px 63.58 % 1 min 1 core @ 2.5 Ghz (C/C++)
J. Kostkova: Stratified dense matching for stereopsis in complex scenes. BMVC 2003.
25 GCSF 12.06 % 13.26 % 1.9 px 2.1 px 60.77 % 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.
26 OCV-BM-post 12.20 % 13.68 % 2.1 px 2.3 px 47.11 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
G. Bradski: The OpenCV Library. Dr. Dobb's Journal of Software Tools 2000.
27 GCS 13.37 % 14.54 % 2.1 px 2.3 px 51.06 % 2.2 s 1 core @ 2.5 Ghz (C/C++)
J. Cech and R. Sara: Efficient Sampling of Disparity Space for Fast And Accurate Matching. BenCOS 2007.
28 CostFilter 19.96 % 21.05 % 5.0 px 5.4 px 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.
29 OCV-BM 25.39 % 26.72 % 7.6 px 7.9 px 55.84 % 0.1 s 1 core @ 2.5 Ghz (C/C++)
G. Bradski: The OpenCV Library. Dr. Dobb's Journal of Software Tools 2000.
30 GC+occ 33.50 % 34.74 % 8.6 px 9.2 px 87.57 % 6 min 1 core @ 2.5 Ghz (C/C++)
V. Kolmogorov and R. Zabih: Computing Visual Correspondence with Occlusions using Graph Cuts. ICCV 2001.
This table as LaTeX

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.



eXTReMe Tracker