The KITTI Vision Benchmark Suite之Stereo Evaluation 2012

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.

• Download stereo/optical flow data set (2 GB)
• Download stereo/optical flow calibration files (1 MB)
• Download multi-view extension (20 frames per scene, all cameras) (17 GB)
• Semantic and instance labels for 60 images and car labels for all training images (1 MB)
• Download stereo/optical flow development kit (3 MB)

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 beeninterpolated 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 innon-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: On 04.11.2013 we have improved theground truth disparity maps andflow fields leading to slightly improvements for all methods. Please download the stereo/flow dataset with theimproved ground truth for training again, if you have downloaded the dataset prior to 04.11.2013. Please consider reporting these new number for all future submissions. Links to lastleaderboards排行榜 before the updates: stereo and flow!

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 usesepipolar geometry核面几何学for computing optical flow
•  Additional training data: Use of additional data sources for training (see details)

Table   AllReflective        Error threshold  2 pixels3 pixels4 pixels5 pixels        Evaluation area  All pixels Estimated pixels 

 MethodSettingCodeOut-NocOut-AllAvg-NocAvg-AllDensityRuntimeEnvironment1GC-NET 1.77 %2.30 %0.6 px0.7 px100.00 %0.9 sNvidia GTX Titan XA. 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. arXiv preprint arxiv:1703.04309 2017.2L-ResMatch code2.27 %3.40 %0.7 px1.0 px100.00 %48 sTitan X (Torch7, CUDA)A. Shaked and L. Wolf: Improved Stereo Matching with Constant Highway Networks and Reflective Loss. arXiv preprint arxiv:1701.00165 2016.3CNNF+SGM  2.28 %3.48 %0.7 px0.9 px100.00 %71 sTESLA K40C 4SN  2.29 %3.50 %0.7 px0.9 px100.00 %67 sTitan X 5PBCP  2.36 %3.45 %0.7 px0.9 px100.00 %68 sNvidia GTX Titan XA. Seki and M. Pollefeys: Patch Based Confidence Prediction for Dense Disparity Map. British Machine Vision Conference (BMVC) 2016.6Displets v2 code2.37 %3.09 %0.7 px0.8 px100.00 %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) 2015.7MC-CNN-acrt code2.43 %3.63 %0.7 px0.9 px100.00 %67 sNvidia 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 .8cfusion

code2.46 %2.69 %0.8 px0.8 px99.93 %70 sGPU (Matlab + CUDA)V. Ntouskos and F. Pirri: Confidence driven TGV fusion. arXiv preprint arXiv:1603.09302 2016.9Displets code2.47 %3.27 %0.7 px0.9 px100.00 %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) 2015.10MC-CNN  2.61 %3.84 %0.8 px1.0 px100.00 %100 sNvidia GTX Titan (CUDA, Lua/Torch7)J. Zbontar and Y. LeCun: Computing the Stereo Matching Cost with a Convolutional Neural Network. Conference on Computer Vision and Pattern Recognition (CVPR) 2015.11PRSM

code2.78 %3.00 %0.7 px0.7 px100.00 %300 s1 core @ 2.5 Ghz (C/C++)C. Vogel, K. Schindler and S. Roth: 3D Scene Flow Estimation with a Piecewise Rigid Scene Model. ijcv 2015.12SPS-StFl

 2.83 %3.64 %0.8 px0.9 px100.00 %35 s1 core @ 3.5 Ghz (C/C++)K. Yamaguchi, D. McAllester and R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014.13MC-CNN-SS  3.02 %4.45 %0.8 px1.0 px100.00 %1.35 s1 core 2.5 Ghz + K40 NVIDIA, Lua-Torch 14VC-SF

 3.05 %3.31 %0.8 px0.8 px100.00 %300 s1 core @ 2.5 Ghz (C/C++)C. Vogel, S. Roth and K. Schindler: View-Consistent 3D Scene Flow Estimation over Multiple Frames. Proceedings of European Conference on Computer Vision. Lecture Notes in, Computer Science 2014.15Content-CNN  3.07 %4.29 %0.8 px1.0 px100.00 %0.7 sNvidia GTX Titan X (Torch)W. Luo, A. Schwing and R. Urtasun: Efficient Deep Learning for Stereo Matching. CVPR 2016.16TBR  3.09 %4.29 %0.7 px0.9 px100.00 %1750 s4 cores @ 3.0 Ghz (Matlab + C/C++) 17Deep Embed  3.10 %4.24 %0.9 px1.1 px100.00 %3 s1 core @ 2.5 Ghz (C/C++)Z. Chen, X. Sun, Y. Yu, L. Wang and C. Huang: A Deep Visual Correspondence Embedding Model for Stereo Matching Costs. ICCV 2015.18JSOSM  3.15 %3.94 %0.8 px0.9 px100.00 %105 s8 cores @ 2.5 Ghz (C/C++)X. Li and J. Liu: EFFICIENT STEREO MATCHING USING SEGMENT OPTIMIZATION. ICIP 2016.19LPU  3.22 %4.27 %0.8 px1.0 px100.00 %1650 s1 core @ 2.5 Ghz (C/C++) 20OSF

code3.28 %4.07 %0.8 px0.9 px99.98 %50 min1 core @ 3.0 Ghz (Matlab + C/C++)M. Menze and A. Geiger: Object Scene Flow for Autonomous Vehicles. Conference on Computer Vision and Pattern Recognition (CVPR) 2015.21CoR code3.30 %4.10 %0.8 px0.9 px100.00 %6 s6 cores @ 3.3 Ghz (Matlab + C/C++)A. Chakrabarti, Y. Xiong, S. Gortler and T. Zickler: Low-level Vision by Consensus in a Spatial Hierarchy of Regions. CVPR 2015.22TCD-CRF  3.32 %5.24 %0.9 px1.9 px100.00 %60 s4 cores @ 3.5 Ghz (C/C++)S. Arjomand Bigdeli, G. Budweiser and M. Zwicker: Temporally Coherent Disparity Maps Using CRFs with Fast 4D Filtering. Proc. ACPR 2015.23SPS-St code3.39 %4.41 %0.9 px1.0 px100.00 %2 s1 core @ 3.5 Ghz (C/C++)K. Yamaguchi, D. McAllester and R. Urtasun: Efficient Joint Segmentation, Occlusion Labeling, Stereo and Flow Estimation. ECCV 2014.24PCBP-SS  3.40 %4.72 %0.8 px1.0 px100.00 %5 min4 cores @ 2.5 Ghz (Matlab + C/C++)K. Yamaguchi, D. McAllester and R. Urtasun: Robust Monocular Epipolar Flow Estimation. CVPR 2013.25STD  3.48 %4.24 %0.9 px1.0 px99.99 %40 min2 cores @ 3.0 Ghz (C/C++) 26MN  3.58 %4.77 %0.9 px1.1 px100.00 %3 min>8 cores @ 2.5 Ghz (C/C++) 27CPM2 code3.58 %4.41 %0.9 px1.1 px99.99 %1.8 s1 core @ 2.5 Ghz (C/C++) 28DDS-SS  3.83 %4.59 %0.9 px1.0 px100.00 %1 min1 core @ 2.5 Ghz (Matlab + C/C++)D. Wei, C. Liu and W. Freeman: A Data-driven Regularization Model for Stereo and Flow. 3DTV-Conference, 2014 International Conference on 2014.29StereoSLIC  3.92 %5.11 %0.9 px1.0 px99.89 %2.3 s1 core @ 3.0 Ghz (C/C++)K. Yamaguchi, D. McAllester and R. Urtasun: Robust Monocular Epipolar Flow Estimation. CVPR 2013.30SMCM  3.94 %5.24 %0.9 px1.1 px100.00 %1800 sNvidia GTX 1080 (Caffe)M. Yang, Y. Liu, Y. Cai and Z. You: Stereo matching based on classification of materials. Neurocomputing 2016.31PR-Sf+E

 4.02 %4.87 %0.9 px1.0 px100.00 %200 s4 cores @ 3.0 Ghz (Matlab + C/C++)C. Vogel, K. Schindler and S. Roth: Piecewise Rigid Scene Flow. International Conference on Computer Vision (ICCV) 2013.32PCBP  4.04 %5.37 %0.9 px1.1 px100.00 %5 min4 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 2012.33DispNetC code4.11 %4.65 %0.9 px1.0 px100.00 %0.06 sNvidia 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 2016.34CSPMS  4.13 %5.92 %1.2 px1.6 px100.00 %6 s4 cores @ 2.5 Ghz (C/C++)J. Cho and M. Humenberger: Fast PatchMatch Stereo Matching Using Multi-Scale Cost Fusion for Automotive Applications. IV 2015.35SGM-post  4.27 %5.33 %1.0 px1.1 px100.00 %5 s4 cores @ 2.5 Ghz (C/C++)Z. Zhong: Efficient Learning based Semi-Global Stereo Matching. 2015 submitted.36SD4CNN  4.27 %5.45 %1.0 px1.1 px100.00 %2 s4 cores @ >3.5 Ghz (C/C++) 37MBM  4.35 %5.43 %1.0 px1.1 px100.00 %0.2 s1 core @ 3.0 Ghz (C/C++)N. Einecke and J. Eggert: A Multi-Block-Matching Approach for Stereo. IV 2015.38PR-Sceneflow

 4.36 %5.22 %0.9 px1.1 px100.00 %150 sec4 core @ 3.0 Ghz (Matlab - C/C++)C. Vogel, K. Schindler and S. Roth: Piecewise Rigid Scene Flow. International Conference on Computer Vision (ICCV) 2013.39CoR-Conf code4.49 %5.26 %1.0 px1.2 px96.37 %6 s6 cores @ 3.3 Ghz (Matlab + C/C++)A. Chakrabarti, Y. Xiong, S. Gortler and T. Zickler: Low-level Vision by Consensus in a Spatial Hierarchy of Regions. CVPR 2015.40sgm  4.50 %5.74 %1.1 px1.3 px96.89 %1 s1 core @ 2.5 Ghz (C/C++) 41TVTGV  4.60 %6.09 %1.3 px1.5 px100.00 %5.6 sGPU @ 2.5 Ghz (C/C++) 42AEGF  4.81 %6.12 %1.2 px1.8 px99.99 %8 s1 core @ 2.5 Ghz (C/C++) 43AARBM  4.86 %5.94 %1.0 px1.2 px100.00 %0.25 s1 core @ 3.0 Ghz (C/C++)N. Einecke and J. Eggert: Block-Matching Stereo with Relaxed Fronto-Parallel Assumption. IV 2014.44wSGM  4.97 %6.18 %1.3 px1.6 px97.03 %6s1 core @ 3.5 Ghz (C/C++)R. Spangenberg, T. Langner and R. Rojas: Weighted Semi-Global Matching and Center-Symmetric Census Transform for Robust Driver Assistance. CAIP 2013.45AABM  4.97 %6.04 %1.0 px1.2 px100.00 %0.12 s1 core @ 3.1 Ghz (C/C++)N. Einecke and J. Eggert: Stereo Image Warping for Improved Depth Estimation of Road Surfaces. IV 2013.46ATGV  5.02 %6.88 %1.0 px1.6 px100.00 %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 2013.47rSGM code5.03 %6.60 %1.1 px1.5 px97.22 %0.2 s4 cores @ 2.6 Ghz (C/C++)R. Spangenberg, T. Langner, S. Adfeldt and R. Rojas: Large Scale Semi-Global Matching on the CPU. IV 2014.48DeepCostAggr  5.08 %6.37 %1.1 px1.4 px100.00 %0.03 sGPU @ 2.5 Ghz (C/C++) 49iSGM  5.11 %7.15 %1.2 px2.1 px94.70 %8 s2 cores @ 2.5 Ghz (C/C++)S. Hermann and R. Klette: Iterative Semi-Global Matching for Robust Driver Assistance Systems. ACCV 2012.50RBM  5.18 %6.21 %1.1 px1.3 px100.00 %0.2 s1 core @ 3.0 Ghz (C/C++)N. Einecke and J. Eggert: Block-Matching Stereo with Relaxed Fronto-Parallel Assumption. IV 2014.51ARW code5.20 %6.87 %1.2 px1.5 px99.33 %4.6s1 core @ 3.5 Ghz (MATLAB+C/C++)S. Lee, J. Lee, J. Lim and I. Suh: Robust Stereo Matching using Adaptive Random Walk with Restart Algorithm. Image and vision computing (accepted) 2015.52DLP  5.28 %7.21 %1.2 px2.0 px100.00 %60 s8 cores @ >3.5 Ghz (C/C++)V. Nguyen, H. Nguyen and J. Jeon: Robust Stereo Data Cost With a Learning Strategy. IEEE Transactions on Intelligent Transportation Systems 2017.53Ensemble  5.34 %6.91 %1.5 px2.0 px100.00 %135 s2 cores @ >3.5 Ghz (Matlab)A. Spyropoulos and P. Mordohai: Ensemble Classifier for Combining Stereo Matching Algorithms. International Conference on 3D Vision (3DV) 2015.54ALTGV  5.36 %6.49 %1.1 px1.2 px100.00 %20 sGPU @ 2.5 Ghz (C/C++)G. Kuschk and D. Cremers: Fast and Accurate Large-scale Stereo Reconstruction using Variational Methods. ICCV Workshop on Big Data in 3D Computer Vision 2013.55SNCC  5.40 %6.44 %1.2 px1.3 px100.00 %0.11 s1 core @ 3.1 Ghz (C/C++)N. Einecke and J. Eggert: A Two-Stage Correlation Method for Stereoscopic Depth Estimation. DICTA 2010.56CAT  5.45 %6.54 %1.1 px1.2 px100.00 %10 s1 core @ 3.5 Ghz (C/C++)J. Ha, J. Jeon, G. Bae, S. Jo and H. Jeong: Cost Aggregation Table: Cost Aggregation Method Using Summed Area Table Scheme for Dense Stereo Correspondence. Advances in Visual Computing 2014.57SGM  5.76 %7.00 %1.2 px1.3 px85.80 %3.7 s1 core @ 3.0 Ghz (C/C++)H. Hirschmueller: Stereo Processing by Semi-Global Matching and Mutual Information. PAMI 2008.58mSGM-LDE  6.01 %8.22 %1.4 px2.4 px100.00 %55 s2 cores @ 2.5 Ghz (C/C++)V. Nguyen, D. Nguyen, S. Lee and J. Jeon: Local Density Encoding for Robust Stereo Matching. TCSVT 2014.59Toast2

 6.16 %7.42 %1.2 px1.4 px95.39 %0.03 s4 cores @ 3.5 Ghz (C/C++)B. Ranft and T. Strau\ss: Modeling Arbitrarily Oriented Slanted Planes for Efficient Stereo Vision based on Block Matching. Intelligent Transportation Systems (ITSC), 2014 IEEE 17th International Conference on 2014.60ITGV  6.20 %7.30 %1.3 px1.5 px100.00 %7 s1 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 2012.61TFS

 6.28 %6.59 %1.4 px1.5 px97.32 %0.4 s1 core @ 3.5 Ghz (C/C++) 62WlinBPM  6.43 %8.60 %1.2 px2.0 px99.99 %2.5 min1 core @ 2.5 Ghz (C/C++)ERROR: Wrong syntax in BIBTEX file.63ELSEP  6.84 %8.02 %1.3 px1.5 px82.28 %1 s1 core @ 2.5 Ghz (Python) 64EGCCS  7.10 %8.58 %1.5 px2.0 px100.00 %280 s1 core @ 3.0 Ghz (C/C++) 65OCV-SGBM code7.64 %9.13 %1.8 px2.0 px86.50 %1.1 s1 core @ 2.5 Ghz (C/C++)H. Hirschmueller: Stereo processing by semiglobal matching and mutual information. PAMI 2008.66SSMW  7.83 %8.95 %1.6 px1.8 px99.99 %2.5 min8 cores @ 2.5 Ghz (C/C++)X. Li, J. Liu, G. Chen and H. Fu: Efficient Methods Using Slanted Support Windows for Slanted Surfaces. IET Computer Vision, http://ietdl.org/t/5QsTxb 2016.67MSMW

code8.01 %9.24 %1.6 px1.7 px72.39 %3 min4 cores @ 2.5 Ghz (C/C++)A. Buades and G. Facciolo: On the performance of local methods for stereovision. 2013 submitted.68ELAS code8.24 %9.96 %1.4 px1.6 px94.55 %0.3 s1 core @ 2.5 Ghz (C/C++)A. Geiger, M. Roser and R. Urtasun: Efficient Large-Scale Stereo Matching. ACCV 2010.69linBP  8.56 %10.70 %1.7 px2.7 px99.89 %1.6 min1 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 2013.70ADSM  8.71 %10.05 %2.1 px2.7 px100.00 %125 s1 core @ 2.0 Ghz (C/C++) 71Deep-Raw  8.93 %11.07 %3.9 px4.9 px100.00 %1 s1 core @ 2.5 Ghz (C/C++)Z. Chen, X. Sun, Y. Yu, L. Wang and C. Huang: A Deep Visual Correspondence Embedding Model for Stereo Matching Costs. ICCV 2015.72S+GF (Cen) code9.03 %11.21 %2.1 px3.4 px100.00 %140 s1 core @ 3.0 Ghz (C/C++)K. Zhang, Y. Fang, D. Min, L. Sun, S. Yang, S. Yan and Q. Tian: Cross-Scale Cost Aggregation for Stereo Matching. CVPR 2014.73CrossCensus  9.46 %10.86 %2.3 px2.7 px100.00 %30 s1 core @ 2.5 Ghz (C/C++)K. Zhang, J. Lu and G. Lafruit: Cross-Based Local Stereo Matching Using Orthogonal Integral Images. Circuits and Systems for Video Technology, IEEE Transactions on 2009.74SymST-GP  9.79 %11.66 %2.5 px3.3 px100.00 %0.254 sDual - Nvidia GTX Titan (CUDA)R. Ralha, G. Falcao, J. Amaro, V. Mota, M. Antunes, J. Barreto and U. Nunes: Parallel refinement of slanted 3D reconstruction using dense stereo induced from symmetry. Journal of Real-Time Image Processing 2016.75SM_GPTM  9.79 %11.38 %2.1 px2.6 px100.00 %6.5 s2 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 2012.76LAMC-DSΜ  9.82 %11.49 %2.1 px2.7 px99.96 %10.8 min2 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 2013.77IIW  10.78 %12.62 %3.3 px4.3 px70.85 %5.5 s1 core @ 2.5 Ghz (C/C++)A. Murarka and N. Einecke: A meta-technique for increasing density of local stereo methods through iterative interpolation and warping. Canadian Conference on Computer and Robot Vision 2014.78SDM code10.95 %12.14 %2.0 px2.3 px63.58 %1 min1 core @ 2.5 Ghz (C/C++)J. Kostkova: Stratified dense matching for stereopsis in complex scenes. BMVC 2003.79HLSC_mesh  11.22 %12.82 %2.3 px2.9 px100.00 %800 s1 core @ 2.5 Ghz (Matlab + C/C++)S. Hadfield, K. Lebeda and R. Bowden: Stereo reconstruction using top-down cues. Computer Vision and Image Understanding 2016.80GF (Census) code11.65 %13.76 %4.5 px5.6 px100.00 %120 s1 core @ 3.0 Ghz (C/C++)A. Hosni, C. Rhemann, M. Bleyer, C. Rother and M. Gelautz: Fast Cost-Volume Filtering for Visual Correspondence and Beyond. TPAMI 2013.
K. Zhang, Y. Fang, D. Min, L. Sun, S. Yang, S. Yan and Q. Tian: Cross-Scale Cost Aggregation for Stereo Matching. CVPR 2014.81BSM code11.74 %13.44 %2.2 px2.8 px97.02 %2.5 min1 core @ 3.0 Ghz (C/C++)K. Zhang, J. Li, Y. Li, W. Hu, L. Sun and S. Yang: Binary stereo matching. Pattern Recognition (ICPR), 2012 21st International Conference on 2012.82GCSF

code12.05 %13.24 %1.9 px2.1 px60.77 %2.4 s1 core @ 2.5 Ghz (C/C++)J. Cech, J. Sanchez-Riera and R. Horaud: Scene Flow Estimation by Growing Correspondence Seeds. CVPR 2011.83OCV-BM-post code12.28 %13.76 %2.1 px2.3 px47.11 %0.1 s1 core @ 2.5 Ghz (C/C++)G. Bradski: The OpenCV Library. Dr. Dobb's Journal of Software Tools 2000.84GCS code13.38 %14.54 %2.1 px2.3 px51.06 %2.2 s1 core @ 2.5 Ghz (C/C++)J. Cech and R. Sara: Efficient Sampling of Disparity Space for Fast And Accurate Matching. BenCOS 2007.85CostFilter code19.99 %21.08 %5.0 px5.4 px100.00 %4 min1 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 2011.86GC+occ code33.49 %34.73 %8.6 px9.2 px87.57 %6 min1 core @ 2.5 Ghz (C/C++)V. Kolmogorov and R. Zabih: Computing Visual Correspondence with Occlusions using Graph Cuts. ICCV 2001.87VariableCros  34.84 %36.11 %12.4 px12.9 px95.66 %30 s1 core @ 2.5 Ghz (Matlab)K. Zhang, J. Lu and G. Lafruit: Cross-Based Local Stereo Matching Using Orthogonal Integral Images. Circuits and Systems for Video Technology, IEEE Transactions on 2009.88ALE-Stereo code50.48 %51.19 %13.0 px13.5 px100.00 %50 min1 core @ 3.0 Ghz (C/C++)L. Ladicky, P. Sturgess, C. Russell, S. Sengupta, Y. Bastanlar, W. Clocksin and P. Torr: Joint Optimisation for Object Class Segmentation and Dense Stereo Reconstruction. BMVC 2010.89MEDIAN  52.61 %53.67 %7.7 px8.2 px99.95 %0.01 s1 core @ 2.5 Ghz (C/C++) 90AVERAGE  61.62 %62.49 %8.0 px8.6 px99.95 %0.01 s1 core @ 2.5 Ghz (C/C++) 

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.

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}
}