Mouseover the table cells to see the produced
disparity map. Clicking a cell will blink the ground truth for
comparison. To change the table type, click the links below.
For more information, please see the description of new features.
The proposed IGEV-Stereo builds a combined
geometry encoding volume that encodes geometry and context information as well as local matching details, and iteratively indexes it to update the disparity map.
We introduce RAFT-Stereo, a new deep architecture for rectified stereo based on the optical flow network RAFT. We introduce multi-level convolutional GRUs, which more efficiently propagate information across the image. A modified version of RAFT-Stereo can perform accurate real-time inference.
Junhong Min and Youngpil Jeon. Confidence aware stereo matching for realistic cluttered scenario. ICIP 2024 submission 1861.
Our approach estimates disparities using implicitly inferred confidence levels. This capability is facilitated by our newly developed U-net transformer, which incorporates various attention mechanisms to extract global and local contexts from rectified image pairs.
feature dimension:128
refinement iterations:3
Intel i9 13900k, Nvidia 4090
02/05/24
231
CAS
F
3
17.1
22
11.2
21
10.8
41
11.1
25
13.4
20
20.7
12
12.5
18
17.7
14
16.9
13
10.3
33
35.2
84
27.5
14
18.3
25
25.8
26
19.5
29
20.1
28
Anonymous. Local expansion moves for stereo matching based on RANSAC confidence. ICCV 2021 submission 3073.
A stereo matching algorithm based on collaborative optimization among pixels is proposed. Based on local expansion, the matching energy function of pixels is defined by using the color and gradient features of adjacent pixels, and the cooperative competition mechanism between pixels is introduced.
iterations = 5; pmIterations = 2;
C/C++,i7-4790 CPU@3.60GHz.
03/05/21
143
LocalExp-RC
H
2
20.0
23
11.9
28
11.9
56
12.6
31
15.0
29
31.4
47
21.4
38
18.1
16
22.4
25
10.5
37
24.0
31
28.6
18
20.1
30
35.1
89
24.3
46
22.6
32
Jie Li, Penglei Ji, and Xinguo Liu. Superpixel alpha-expansion and normal adjustment for stereo matching. Proceeding of CAD/Graphics 2019.
Junda Cheng and Gangwei Xu. CoAtRS stereo: Fully exploiting convolution and attention for stereo matching. Submitted to IEEE Transactions on Multimedia, 2021.
The approach relies on a fast multi-resolution initialization step, differentiable 2D geometric propagation and warping mechanisms to infer slanted plane hypotheses at multiple resolutions.
see paper
Titan V
10/28/20
131
HITNet
H
2
20.7
31
13.2
41
10.8
40
12.7
33
14.4
23
30.2
41
20.8
34
24.2
34
23.4
28
8.85
27
49.3
146
32.5
30
19.6
28
26.2
27
20.5
32
28.3
53
Xianjing Cheng and Yong Zhao. HLocalExp-CM: Confidence map by hierarchical local expansion moves for stereo matching. To appear in Journal of Electronic Imaging, 2022.
We propose local expansion moves for estimating dense 3D labels on a pairwise MRF. The data term uses a PatchMatch-like 3D slanted window formulation, where raw matching costs within a window are computed by MC-CNN-acrt and aggregated using guided image filtering. The smoothness term uses a pairwise curvature regularization term by Olsson et al. 2013.
Haoxuan Sun and Taoyang Wang. Weighted RANSAC disparity refinement based on estimated single-view normal map and SAM. Submitted to IEEE TIP 2023.
Intel(R) Core(TM) i7-10875H CPU @ 2.30GHz
11/12/23
222
SNDR
H
2
21.3
35
16.9
63
13.5
71
13.4
41
14.2
22
27.5
29
24.6
58
27.2
42
23.0
27
20.7
130
24.0
30
28.9
22
27.8
70
30.0
42
16.0
19
15.3
15
Anonymous. Estimate regularization weight for local expansion moves stereo matching. ACPR 2021 submission.
The method that estimate optimal parameters for MRF stereo can not be directly used to estimate parameters for local expansion moves stereo. To estimate regularization weight for local expansion moves stereo, we propose the probabilistic mixture models for slanted patch matching terms and curvature regularization terms.
This paper presents an accurate and efficient hierarchical BP framework using the representation of the image segmentation pyramid (ISP). We design a hierarchy of MRF networks using the graph of superpixels at each ISP level.
We propose a cost aggregation method that efficiently weave together MST-based support region filtering and PatchMatch-based 3D label search. We use the raw matching cost of MC-CNN.
A 3D label based method with global optimization at pixel level. A bilayer matching cost is employed by first matching small square windows then aggregate on large irregular windows. Global optimization is carried out by fusing candidate proposals, which are generated from our specific superpixel structure.
Anonymous. Towards adaptive non-parametric modeling for disparity estimation. CVPR 2023 submission 7806.
RTX3090
11/11/22
197
AnPM
F
3
23.0
46
15.7
56
11.7
51
11.7
27
17.0
41
40.5
89
24.0
54
25.0
38
19.0
18
11.1
44
29.4
67
32.5
29
24.7
48
31.5
52
39.4
116
27.3
48
Xianjing Cheng, Yong Zhao, Zhijun Hu, Xiaomin Yu, Ren Qian, and Haiwei Sang. Superpixel cut-based local expansion for accurate stereo matching. IET Image Processing, 2021.
i7 CPU @2.2GH,C++, 8 cores
04/22/21
144
LESC
H
2
23.1
47
12.8
35
13.2
69
13.5
43
18.0
47
34.9
64
23.0
46
28.2
44
25.1
35
12.9
68
26.0
41
37.4
46
24.9
49
36.6
101
24.6
50
26.9
45
Anonymous. An improved RaftStereo trained with multiple mixed datasets for Robust Vision Challenge. RVC 2022 submission.
GTX 2080ti
10/03/22
185
iRaftStereo_RVC
H
2
24.0
48
18.0
71
16.0
110
15.4
60
19.3
54
27.9
31
25.8
66
29.1
46
26.6
45
14.4
80
36.3
91
38.4
49
24.3
47
29.0
37
25.5
58
27.7
52
Gongben Han and Kanjian Zhang. Stereo matching via cost aggregation and iterative optimization. Submitted to IEEE TPAMI, 2023.
NVIDIA RTX 3090 (PyTorch)
12/07/23
226
4D-IteraStereo
H
2
24.2
49
11.4
23
10.3
35
13.2
39
16.8
39
36.1
71
25.9
67
44.0
107
46.0
133
9.51
30
25.4
37
34.8
39
18.1
24
29.4
38
23.5
43
23.3
38
Anonymous. Extracting geometric relations from estimated single-view normal and refining disparity. CVPR 2023 submission 4986.
We design a full-convolutional network to generate disparity map as a regression problem. Applying pyramid pooling and skip connection to integrate hierarchical context information.
We extend the standard BP sequential technique to the fully connected CRF models with the geodesic distance affinity.
Also a new approach to the BP marginal solution is proposed that we call one-view-occlusion detection (OVOD). In contrast to the standard winner takes all (WTA) estimation, the proposed OVOD solution allows to find occluded regions in the disparity map and simultaneously improve the matching result.
As a result we can perform only one energy minimization process and avoid the cost calculation for the second view and the left-right check procedure.
All parameter settings are given in the C++ MS VS project available at the project website.
Intel(R) Xeon(R) CPU E5-1620 v4 @3.50 GHz
12/11/17
56
OVOD
H
2
24.6
52
13.1
38
10.1
32
14.4
51
20.2
61
35.6
67
26.9
72
30.1
50
34.4
70
10.1
31
28.4
61
39.9
53
28.5
72
31.9
61
24.3
46
35.1
80
Peng Yao and Jieqing Feng. Stacking learning with coalesced cost filtering for accurate stereo matching. Submitted to Journal of Visual Communication and Image Representation 2020.
By leveraging Stacking Learning with Coalesced Cost Filtering to make the conventional algorithms achieve more accurate disparity estimations.
For the Random Forest, we set 10 Decision Trees, maximum depth is 25 and minimum number of samples in each node to split equal to 12.
An efficient stereo matching algorithm, which applies adaptive smoothness constraints using texture and edge information, is proposed in this work. First, we determine non-textured regions, on which an input image yields flat pixel values. In the non-textured regions, we penalize depth discontinuity and complement the primary CNN-based matching cost with a color-based cost. Second, by combining two edge maps from the input image and a pre-estimated disparity map, we extract denoised edges that correspond to depth discontinuity with high probabilities. Thus, near the denoised edges, we penalize small differences of neighboring disparities.
The method uses the MC-CNN code for the matching cost computation only.
We propose a feature ensemble network leveraging deep convolutional neural network to perform matching cost computation and the disparity refinement. For matching cost computation, patch-based network architecture with multi-size and multi-layer pooling unit is adopted to learn cross-scale feature representations. For disparity refinement, the initial optimal and sub-optimal disparity maps are incorporated and diverse base learners are applied.
We propose a stereo matching algorithm that directly refines the winner-take-all (WTA) disparity map by exploring its statistic significance. WTA disparity maps are obtained from the pre-computed raw matching costs of MC-CNN-acrt.
A novel pooling scheme is used to train a matching cost function with a CNN. It widens the size of receptive field effectively without losing the fine details.
The overall post-processing pipeline is kept almost same as the original MC-CNN-acrt, except that the parameter settings are changed as follows:
cbca_num_iterations_1 = 0, cbca_num_iterations_2 = 1, sgm_P1 = 1.3, sgm_P2 = 17.0, sgm_Q1 = 3.6, sgm_Q2 = 36.0, and sgm_V = 1.4.
Torch; the Intel core i7 4790K
CPU and a single Nvidia Geforce GTX Titan X GPU
Semi-Global Matching (SGM) uses an aggregation scheme to combine costs from multiple 1D scanline optimizations that tends to hurt its accuracy in difficult scenarios. We propose replacing this aggregation scheme with a new learning-based method that fuses disparity proposals estimated using scanline optimization. Our proposed SGM-Forest algorithm solves this problem using per-pixel classification. SGM-Forest currently ranks 1st on the ETH3D stereo benchmark and is ranked competitively on the Middlebury 2014 and KITTI 2015 benchmarks. It consistently outperforms SGM in challenging settings and under difficult training protocols that demonstrate robust generalization, while adding only a small computational overhead to SGM.
Using MC-CNN-acrt matching cost
Intel Xeon CPU E5-2697, 32GB RAM
03/11/18
61
SGM-Forest
H
2
26.2
60
16.2
59
14.3
85
15.6
61
22.0
75
35.8
68
26.5
69
34.4
64
37.5
87
12.6
58
28.6
62
41.5
64
28.7
73
32.5
66
25.9
60
33.9
75
Anonymous. Cascade and fuse cost volume for efficient and robust stereo matching. CVPR 2021 submission 1728.
we construct multi-scale cost volumes and fuse lower scale cost volumes and cascade higher scale ones to realize efficient and robust stereo matching
we first pre-train our model on sceneflow dataset and then finetune it jointly on Middlebury + KITTI + ETH3D
We propose a method to combine the predicted surface normal constraint by deep learning. With the selected reliable disparities from stereo matching method and effective edge fusion strategy, we can faithfully convert the predicted surface normal map to a disparity map by solving a least squares system which maintains discontinuity. We use the raw matching cost of MC-CNN.
In stereo matching, there are two cases of poor performance: (1) the interior of large objects, and (2) object boundaries and small objects. In this work, we present feature enhancement stereo matching network to solve the problems.
None
2080ti
11/21/21
161
FENet
H
2
26.5
66
23.0
86
9.24
26
11.7
29
18.9
51
49.3
123
22.3
41
49.1
123
47.1
138
10.5
40
27.5
54
35.5
41
25.0
50
29.7
40
22.1
36
35.2
82
Peng Yao, Haiwei Sang, and Xu Cheng. Structured support vector machine with coarse-to-fine PatchMatch filtering for stereo matching. Submitted to The Visual Computer, 2023.
Stereo Matching Using Structured Supported Vector Machine and Coarse to Fine Features
see the paper
i7-8700,C/C++
06/09/23
207
SSVM-CFPMF
H
2
26.7
67
22.4
85
13.8
75
16.2
70
20.8
69
20.8
13
25.4
63
30.2
51
37.5
86
25.2
154
32.4
75
42.2
72
29.7
85
39.2
121
21.2
35
32.1
61
Chao He. Local stereo matching with side window. Submitted to IEEE Signal Processing Letters, 2022.
We propose four efficient feature extractors based on convolutional neural networks for stereo matching cost computation. Two of them generate multiscale features with diverse receptive field sizes. These multiscale features are used to compute the corresponding multiscale matching costs. We then determine an optimal cost by combining the multiscale costs using edge information. On the other hand, the other two feature extractors produce uni-scale features by combining multiscale features directly through fully connected layers. Finally, after obtaining matching costs using one of the four extractors, we determine optimal disparities based on the cross-based cost aggregation and the semiglobal matching.
Compute the matching cost with a convolutional neural network (accurate architecture). Then apply cross-based cost aggregation, semiglobal matching, left-right consistency check, median filter, and a bilateral filter.
DETAILS:
The network is similar to the one described in our CVPR paper differing only in the values of some hyperparameters. The input to the network are two 11 x 11 image patches. Five convolutional layers with 3 x 3 kernels and 112 feature maps extract feature vectors from the input image patches. The two 112-length feature vectors are concatenated into a 224-length vector which is passed through three fully-connected layers with 384 units each. The final (fourth) fully-connected layer projects the output to a single number---the matching cost. One important addition was the use of data augmentation techniques to increase the size of the training set. We tried to use as much training data as possible. Therefore we combined all of the 2001, 2003, 2005, 2006, and 2014 Middlebury datasets obtaining 60 image pairs. For the newer datasets (2005, 2006, and 2014) we also used several illumination and exposure settings.
See paper.
NVIDIA GTX TITAN Black
08/28/15
20
MC-CNN-acrt
H
2
27.3
75
17.0
64
14.9
94
16.9
80
20.4
63
34.0
58
28.4
86
38.4
82
39.3
97
12.9
65
29.8
68
43.0
81
30.3
89
32.2
63
28.9
69
33.8
74
Anonymous. Revisiting cost aggregation in stereo matching from disparity classification. CVPR 2023 submission 1116.
Cost aggregation plays a critical role in existing stereo
matching methods. Generally, aggregating matching costs
in homogeneous regions with similar disparities is benefi-
cial to matching accuracy. However, previous approaches
commonly use 3D convolutions for cost aggregation with-
out considering the homogeneity of different regions. In
this paper, we revisit cost aggregation in stereo match-
ing from a perspective of disparity classification and pro-
pose a generic yet efficient Disparity Context Aggregation
(DCA) module to improve the performance of CNN-based
methods.
Parameters:4.96 M;
Only using half-resolution Middlebury training images for validation.
We post-process the depth maps produced by Zbontar & LeCun's MC-CNN technique. We use a domain transform to compute an edge-aware variance measure of our confidence in the depth map, and then run our robust bilateral solver on that depth map and confidence with a Geman-McClure loss function.
The MC-CNN is computed using the publicly-available implementation (https://github.com/jzbontar/mc-cnn) which using the GPU, and the robust bilateral solver is computed using our CPU implementation which does not use the GPU, and is written in vanilla C++.
Intel(R) Xeon(R) CPU E5-1650 0 @ 3.20GHz, 6 cores; 32 GB RAM; NVIDIA GTX TITAN X
11/03/15
24
MC-CNN+RBS
H
2
27.5
77
17.7
69
16.0
112
17.0
83
20.2
62
33.7
56
28.2
83
39.2
86
40.3
107
13.6
73
29.4
66
43.8
84
29.3
80
31.6
54
29.3
73
32.4
64
Tuming Yuan. Hourglass cascaded recurrent stereo matching network. Submitted to Image and Vision Computing, 2024.
combine stacked hourglass modules and
recurrent neural networks
Compute the matching cost with a convolutional neural network (fast architecture). Then apply cross-based cost aggregation, semiglobal matching, left-right consistency check, median filter, and a bilateral filter.
Accurate disparity prediction is a hot spot in computer vision, and how to efficiently exploit contextual information is the key to improve the performance. In this paper, we propose a simple yet effective non-local context attention network (NLCANet) to exploit the global context information by using attention mechanisms and semantic information for stereo matching. First, we develop a 2D geometry feature learning (GFL) module to get a more discriminative representation by taking advantage of multi-scale features and form them into the variance-based cost volume. Then, we construct a non-local attention matching (NLAM) module by using the non-local block and hierarchical 3D convolutions, which can effectively regularize the cost volume and capture the global contextual information. Finally, we adopt a geometry refinement (GR) module to refine the disparity map to further improve the performance. Moreover, we add the warping loss function to help the model learn the matching rule of the non-occluded region. Our experiments show that (1), our approach achieves competitive results on KITTI and SceneFlow datasets in the end-point error (EPE) and the fraction of erroneous pixels (D 1 ); (2), our proposed method particularly has superior performance in the reflective regions and occluded areas.
A robust solution for semi-dense stereo matching is presented. It utilizes two CNN models for computing stereo matching cost and performing confidence-based filtering, respectively. Compared to existing CNNs-based matching cost generation approaches, our method feeds additional global information into the network so that the learned model can better handle challenging cases, such as lighting changes and lack of textures. Through utilizing non-parametric transforms, our method is also more self-reliant than most existing semi-dense stereo approaches, which rely highly on the adjustment of parameters.
Matlab, GTX1080Ti, Lua, Python
06/27/18
76
DCNN
H
2
29.4
87
17.5
67
14.7
90
18.1
92
22.0
77
31.4
48
29.6
96
44.7
111
41.2
111
14.0
76
35.9
89
48.5
107
34.0
105
32.2
65
32.3
85
36.1
85
Krishna Shankar, Mark Tjersland, Jeremy Ma, Kevin Stone, and Max Bajracharya. A learned stereo depth system for robotic manipulation in homes. ICRA 2022 submission.
A lightweight network with dilated ResNet feature extractor, a correlation cost volume run at a low resolution, and a refinement network to get a full resolution disparity output. Sparse disparity is processed from the dense disparity using a threshold on the network confidence output and a region grower to remove suspected bad disparities.
It has been proposed by many researchers that combining deep neural networks with graphical models can create more efficient and better regularized composite models. The main difficulties in implementing this in practice are associated with a discrepancy in suitable learning objectives as well as with the necessity of approximations for the inference. In this work we take one of the simplest inference methods, a truncated max-product Belief Propagation, and add what is necessary to make it a proper component of a deep learning model: We connect it to learning formulations with losses on marginals and compute the backprop operation. This BP-Layer can be used as the final or an intermediate block in convolutional neural networks (CNNs), allowing us to design a hierarchical model composing BP inference and CNNs at different scale levels. The model is applicable to a range of dense prediction problems, is well-trainable and provides parameter-efficient and robust solutions in stereo, optical flow and semantic segmentation.
We propose a robust learning-based method for stereo cost volume computation. We accomplish this by coalescing diverse evidence from a bidirectional matching process via random forest classifiers. We show that our matching volume estimation method achieves similar accuracy to purely data-driven alternatives and that it generalizes to unseen data much better. In fact, we used the same model trained on Middlebury 2014 dataset to submit to the KITTI and ETH3D benchmarks.
This is a new weakly supervised method that allows to learn deep metric for stereo reconstruction from unlabeled stereo images, given coarse information about the scenes and the optical system. The deep metric architecture is similar to MC-CNN fst.
Xianjing Cheng and Yong Zhao. Local PatchMatch based on superpixel cut for efficient high-resolution stereo matching. Submitted to BABT (Brazilian Archives of Biology and Technology), 2021.
we propose an efficient method,i.e, local PatchMatch based on superpixel cut for high-resolution stereo matching.
the number of superpixels N is 500, two iterative parameters: k_fea is set to 9 and k_SP is set to 7. The parameter γ to measure the similarity weight is set to 50 and k=8000.
This approach triangulates the polygonized SLIC segmentations of the input images and optimizes a lower-layer MRF on the resulting set of triangles defined by photo consistency and normal smoothness. The lower-layer MRF is solved by a quadratic relaxation method which iterates between PatchMatch and Cholesky Decomposition. The lower-layer MRF is assisted by a upper-layer MRF defined on the set of triangle vertices which exploits local 'visual complexity' cues and encourages smoothness of the vertices' splitting properties. The two layers interact through an Alignment energy term which requires triangles sharing a non-split vertex to have their disparities agree on that vertex. Optimization of the whole model is iterated between optimizations of the two layers till convergence where the upper-layer can be solved in closed form.
omega=0.2
tau_grad=15
theta goes from 0 to 100 by smoothstep function in ten iterations
gamma1=30
gamma2=60
gamma3=0.8
i7-2600 3.40GHz 8 cores, C++
04/19/15
19
MeshStereo
H
2
32.9
106
18.8
73
16.8
119
23.1
130
33.3
134
30.4
43
32.9
117
40.2
89
37.7
89
19.0
117
44.6
120
50.1
113
33.4
101
43.7
146
39.1
113
39.5
93
Philippe Weinzaepfel, Vincent Leroy, Thomas Lucas, Romain Bregier, Yohann Cabon, Vaibhav Arora, Leonid Antsfeld, Boris Chidlovskii, Gabriela Csurka, and Jerome Revaud. Self-supervised pretraining for 3D vision tasks by cross-view completion. NeurIPS 2022; RVC 2022 submission.
Junhong Min and Youngpil Jeon. Confidence-aware symmetric stereo matching via u-net transformer. Submitted to ICRA 2024.
We propose a novel deep stereo matching network a new real-world stereo dataset of cluttered objects taken with a commercially available stereo sensor. We design a U-shaped architecture with various types of attentions which more efficiently extracts global and local contexts from rectified image pairs, resulting in highly accurate disparities. Furthermore, its symmetric structure allows simultaneous estimation both left and right disparity. It can also implicitly estimate the uncertainty i.e. the confidence of estimated disparities.
4 level unet for feature extraction
and 3 level unet for refinement
channel dimension is 128.
I9 13900k, Nvidia RTX 4090
09/14/23
214
CASS
F
3
33.0
109
19.6
76
17.9
126
27.6
155
26.7
100
42.1
93
22.9
44
33.5
61
36.4
80
24.3
149
54.6
154
46.2
98
34.5
109
39.5
123
45.4
143
45.8
114
Yang Xiaowei and Feng Zhiguo. Attention guide cost volume for stereo matching. Submitted to IET Image Processing, 2022.
GTX 3090
10/06/22
186
AGCVNet
H
2
33.1
110
28.1
103
12.9
64
17.8
90
27.6
108
43.9
100
34.0
133
50.8
129
43.8
121
22.4
137
39.4
102
48.2
105
36.2
123
33.3
74
36.0
98
39.5
94
Zhien Dai and Zhaohui Tang. KPEA-Stereo: An adaptive stereo matching method. Submitted to IEEE Transactions on Instrumentation and Measurement, 2022.
We proposed a disparity estimation network named KPEANet.
The proposed model is trained by half-resolution Middlebury and Sceneflow datasets.
999
12/12/22
200
KPEA-Stereo
H
2
33.4
111
29.1
105
15.5
102
19.1
98
26.3
99
41.5
92
27.7
79
49.4
124
40.1
104
22.1
135
40.7
107
48.4
106
33.0
99
36.5
100
39.1
112
57.1
142
Han Li. Adaptive slice stereo matching network. Submitted to Image and Vision Computing, 2022.
Anonymous. Semi-synthesis: a fast way to produce effective datasets for stereo matching. CVPR 2021 submission 3688.
We propose a novel method namely semi-synthesis for producing large-scale on demand stereo datasets which doesn't require further fine-tuning on real datasets, i,e, we haven't fine-tuned the submission model on Middlebury training data.
This paper proposes a new image-guided non-local dense matching method with a three-step optimization based on the combination of image-guided methods and energy function-guided methods.
Cost Computation:
Window Size: 5
Weighting Coefficient: 0.3
Truncation Threshold (Census): 15
Truncation Threshold (HOG): 1
Image-guided Non-local Matching:
Smooth Term: 6
Penalty Term P1: 0.3
Penalty Term P2: 3
Disparity Interpolation:
Truncation Threshold: 5
Smooth Term: 3
Penalty Term P1: 3
Penalty Term P2: 30
Function Base: 5
C/C++ 1 i7 core @3.2 GHz
12/18/15
25
INTS
H
2
34.8
117
40.7
134
16.8
118
22.9
129
31.1
127
54.0
133
32.7
112
45.4
114
38.9
93
17.9
107
44.1
118
50.9
117
36.0
122
40.9
129
33.6
91
48.6
122
Anonymous. An adaptive multi-modal cross-entropy loss for stereo matching. ICRA 2023 submission 1024.
A fast method for high-resolution stereo matching without exploring the full search space. Plane hypotheses are generated from sparse feature matches. Around each plane, a local plane sweep with +/- 3 disparities levels is performed to establish local disparity hypotheses via SGM using NCC matching costs. Finally, each pixel is assigned to one hypothesis using global optimization, again using SGM.
nRounds=3
The full set of parameters is listed in the paper and the supplemental materials on the project webpage.
Hierarchical MGM-16 where coarser level results limit per pixel disparity search range. Post-Processing at each level include Joint Bilateral Filter, Peak removal and, consistency check. The final disparity maps are interpolated using Discontinuity preserving interpolation
See Paper
C/C++; Intel(R) Xeon(R) Gold 6154 CPU @ 3.00GHz, 16 Cores
Efficient two-pass aggregation with census/gradient cost metric, followed by iterative cost penalization and disparity re-selection to encourage local smoothness of disparities.
census window size = 9 x 7
max census distance = 38.03
max gradient difference = 2.51
census/gradient balance = 0.09
aggregation window size = 33 x 33
aggregation range parameter = 23.39
aggregation spatial parameter = 7.69
refinement window size = 65 x 65
refinement range parameter = 11.30
refinement spatial parameter = 17.20
cost penalty coefficient = 0.0023
median filter window size = 3 x 3
3 iterations of refinement
confidence threshold of 0.1 for sparse maps
CUDA C++, NVIDIA GeForce TITAN Black
10/07/14
14
IDR
H
2
36.4
123
56.1
171
11.8
53
19.3
102
37.7
149
57.6
141
34.2
137
55.4
140
40.1
105
17.0
100
45.9
129
52.2
127
36.0
121
39.0
118
39.5
117
39.4
92
Yang Zhang, Peng Song, and Bo Song. A local side window algorithm with tree segmentation for stereo matching. Submitted to Laser and Optoelectronics Progress 2023.
This model is trained on low-resolution data but aims at high-resolution images. It uses a recurrent module to iteratively update a coarse disparity prediction. Then a special refinement module makes a final adjustment. The recurrent update and final refine are applied in a patch-wise manner across the initial disparity.
Trained on Scene Flow, Middlebury 1/4 size, and TartanAir (sampled) datasets. Training disparity range 256 pixels, testing range over 1000 pixels.
Trained on 4 Tesla V100 GPUs. Inference on 1 Tesla V100 GPU.
03/05/21
142
ORStereo
F
3
37.2
126
60.6
191
19.6
136
21.2
118
37.1
145
59.6
146
33.5
125
50.0
125
30.4
58
16.4
92
49.1
144
58.4
157
38.5
132
40.9
128
40.5
122
47.0
119
Xue Liu. Stereo matching with monocular augmentation. Submitted to Signal Processing Letters, 2022.
The method generates multiple proposals on absolute and relative disparities from multi-segmentations. The proposals are coordinated by point-wise competition and pairwise collaboration within a MRF model. During inference, a dynamic programming is performed in different directions with various step sizes.
In this work, we propose a learning-based method to denoise and refine disparity maps of a given stereo method. The proposed variational network arises naturally from unrolling the iterates of a proximal gradient method applied to a variational energy defined in a joint disparity, color, and confidence image space. Our method allows to learn a robust collaborative regularizer leveraging the joint statistics of the color image, the confidence map and the disparity map. Due to the variational structure of our method, the individual steps can be easily visualized, thus enabling interpretability of the method. We can therefore provide interesting insights into how our method refines and denoises disparity maps. The efficiency of our method is demonstrated by the publicly available stereo benchmarks Middlebury 2014 and Kitti 2015.
GTX 2080Ti
05/13/19
94
VN
H
2
37.5
129
23.6
89
23.6
154
28.6
159
29.9
121
39.3
81
38.5
161
42.7
103
42.9
119
27.0
165
48.0
139
54.0
133
43.7
165
50.2
177
38.7
110
40.9
98
Anonymous. UnDAF: A general unsupervised domain adaptation framework for disparity, optical flow or scene flow estimation. CVPR 2021 submission 236.
Pytorch
11/12/20
134
UnDAF-GANet
H
2
37.5
130
12.9
37
11.8
52
34.9
181
42.5
170
49.7
124
46.6
186
50.1
126
46.5
136
10.4
35
59.9
163
56.9
150
26.2
56
50.9
180
28.9
67
61.4
159
Pengxiang Li, Chengtang Yao, Yunde Jia, and Yuwei Wu. Inter-scale similarity guided cost aggregation for stereo matching. Submitted to IEEE Transactions on Circuits and Systems for Video Technology, 2022.
The images are Census transformed and the Hamming distance is used as pixelwise matching cost. Aggregation is performed by a kind of dynamic programming along 8 paths that go from all directions through the image. Small disparity patches are invalidated. Interpolation is also performed along 8 paths.
A fast method for high-resolution stereo matching without exploring the full search space. Plane hypotheses are generated from sparse feature matches. Around each plane, a local plane sweep with +/- 3 disparities levels is performed to establish local disparity hypotheses via SGM using NCC matching costs. Finally, each pixel is assigned to one hypothesis using global optimization, again using SGM.
nRounds=3
The full set of parameters is listed in the paper and the supplemental materials on the project webpage.
Block-matching stereo with Summed Normalized Cross-Correlation (SNCC) measure. Standard post-processed is applied, including a left-right check, error island removal (region growing), hole-filling and median filtering.
SNCC (first stage 3x3, second stage 11x11)
min correlation threshold = 0.3
region growing threshold = 2.5 disparity
min region size = 200 pixel
median filter = 1x5 and 5x1
We propose a novel method for stereo estimation, combining advantages of convolutional neural networks (CNNs) and optimization-based approaches. The optimization, posed as a conditional random field (CRF), takes local matching costs and consistency-enforcing (smoothness) costs as inputs, both estimated by CNN blocks. To perform the inference in the CRF we use an approach based on linear programming relaxation with a fixed number of iterations. We address the challenging problem of training this hybrid model end-to-end. We show that in the discriminative formulation (structured support vector machine) the training is practically feasible. The trained hybrid model with shallow CNNs is comparable to state-of-the-art deep models in both time and performance. The optimization part efficiently replaces sophisticated and not jointly trainable (but commonly applied) post-processing steps by a trainable, well-understood model.
-
NVidia Titan X
03/22/17
48
JMR
H
2
39.4
143
20.9
83
20.9
145
29.7
162
39.0
156
48.1
119
38.8
162
59.4
163
59.1
185
19.2
123
38.3
97
50.4
114
37.2
127
52.7
188
35.0
96
46.5
116
Wang Yun and Wang Longguang. ADStereo: Learning stereo matching from adaptive downsample with disparity alignment. Submitted to IEEE TIP, 2023.
The project proposes a stereo matching network based on neural operator, which can achieve mapping from RGB image pair space to disparity space. This network supports users to test images at any scale, and can customize the disparity range according to different scenarios, and dynamically build Cost Volume based on different scales and disparity ranges.
parser.add_argument('--model', default='gwcnet-g', help='select a model structure', choices=__models__.keys())
parser.add_argument('--maxdisp', type=int, default=192, help='maximum disparity')
parser.add_argument('--start_disp', type=int, default=15, help='maximum disparity')
parser.add_argument('--end_disp', type=int, default=303, help='maximum disparity')
parser.add_argument('--dataset', required=True, help='dataset name', choices=__datasets__.keys())
parser.add_argument('--datapath', required=True, help='data path')
parser.add_argument('--testlist', required=True, help='testing list')
parser.add_argument('--lr', type=float, default=0.001, help='base learning rate')
parser.add_argument('--test_batch_size', type=int, default=1, help='testing batch size')
parser.add_argument('--epochs', type=int, required=True, help='number of epochs to train')
parser.add_argument('--lrepochs', type=str, required=True, help='the epochs to decay lr: the downscale rate')
parser.add_argument('--logdir', required=True, help='the directory to save logs and checkpoints')
parser.add_argument('--loadckpt', help='load the weights from a specific checkpoint')
parser.add_argument('--resume', action='store_true', help='continue training the model')
parser.add_argument('--seed', type=int, default=1, metavar='S', help='random seed (default: 1)')
parser.add_argument('--summary_freq', type=int, default=20, help='the frequency of saving summary')
parser.add_argument('--save_freq', type=int, default=1, help='the frequency of saving checkpoint')
parser.add_argument('--out_add', type=str)
parser.add_argument('--key_query_same', type=str)
parser.add_argument('--deformable_groups', type=int, required=True)
parser.add_argument('--output_representation', type=str, required=True, help='regressing disparity')
parser.add_argument('--sampling', type=str, default='dda', required=True)
parser.add_argument('--scale_min', type=float, default=1)
parser.add_argument('--scale_max', type=float, default=1)
python; 4 Tesla V100 GPU
02/20/24
233
DispNO
H
2
40.2
146
42.0
139
18.1
128
24.2
137
35.5
141
66.3
166
37.0
158
50.7
128
46.2
134
32.5
179
58.7
160
51.9
126
40.2
145
39.3
122
46.8
147
46.4
115
James Okae, Juan Du, and Yueming Hu. Robust statistical approach to stereo disparity map denoising and refinement. Submitted to Journal of Control Theory and Technology, 2020.
Using robust statistics and probability to detect and refine outliers in disparity maps by leveraging the joint statistics of the given disparity map and its reference image.
lamda=1,r1=5,r2=25, sigma=10,tho_d=1, tho_s=4
Matlab Intel® Core™ i7-4600U CPU
05/14/20
112
SRM
H
2
40.6
147
33.1
116
31.0
185
30.4
165
38.0
151
48.1
119
32.2
111
54.1
136
54.5
164
25.5
155
45.2
124
52.6
128
40.4
148
45.5
156
41.1
126
50.9
129
Anonymous. DualNet: Self-supervised stereo based on knowledge distillation. ECCV 2024 submission 327.
Unsupervised Stereo Matching methods have made significant strides recently. However, these approaches have predominantly relied on the assumption of photometric consistency, leading to potential limitations: sensitivity to illuminance changes and difficulty in dealing with problematic areas like occluded or textureless regions.
To mitigate these limitations, this paper introduces a novel self-supervised dual-level framework named \textbf{\textit{Dual-Net}}.
This framework mainly consists of two key components: self-supervised teacher training and student training based on knowledge distillation.
Specifically, the teacher model is first trained in a self-supervised fashion with a focus on feature space and data augmentation consistency.
On the one hand, pixels from feature space are robust to noise and luminance changes, which are discriminative even in textureless regions.
On the other hand, a data augmentation consistency loss is presented to guide the model toward enhanced contextual awareness, thus leading to a completed depth estimation in problematic regions.
Then, the knowledge learned by the teacher model is distilled and transferred probabilistically to the student model. By leveraging this distilled knowledge, the student model is guided by validated insights, enabling it to outperform its teacher model by a large margin.
A modification of the FlowNet 2 architecture [1] for the Robust Vision 2018 Stereo Challenge.
[1] E. Ilg, N. Mayer, T. Saikia, M. Keuper, A. Dosovitskiy, and T. Brox. Flownet 2.0: Evolution of optical flow estimation with deep networks. CVPR 2017.
The images are Census transformed and the Hamming distance is used as pixelwise matching cost. Aggregation is performed by a kind of dynamic programming along 8 paths that go from all directions through the image. Small disparity patches are invalidated. Interpolation is also performed along 8 paths.
We propose a novel lightweight network for stereo estimation. The method uses densely connected layer structures to learn expressive features without the need of fully-connected layers or 3D convolutions. This leads to a network structure with only 0.37M parameters while still having competitive results. The post-processing consists of filtering, a consistency check and hole filling.
GA-Net reference submission as baseline for the stereo benchmark of the robust vision challenge 2020.
All method credits go to the original author (Zhang et al.)
Submission by Nicolas Jourdan, TU Darmstadt, RVC 2020 team.
Trained on Middleburry, KITTI, ETH3D from the KITTI checkpoint made available in the GANet repository on Github by the original authors.
Frequency of sampling was adapted to the dataset size. Test images scaled to next multiple of 48.
This paper presents a novel unsupervised stereo matching cost for stereo matching. Specifically, a novel two-branch convolutional sparse coding (CSC) is used to learn the convolution filter bank without ground truth disparity maps. Then, the sparse representations over the learned convolutional filter bank are utilized to measure the similarity between image patches, namely, the stereo matching cost can be computed by measuring the l1 distance between sparse representations of image patches.
Matlab/C/C++; 1 i5 core @2.9 GHz
04/12/19
92
TCSCSM
H
2
43.3
156
64.7
200
24.9
160
31.1
168
45.3
180
63.8
158
35.8
149
55.8
142
45.3
129
26.5
161
49.6
150
56.3
146
45.6
177
46.4
165
44.8
139
43.9
108
OpenCV 2.4.8 StereoSGBM method, single-pass variant. Reimplementation and modification of H. Hirschmüller's SGM method (CVPR 2006; PAMI 2008).
OpenCV's "semi-global block matching" method; memory efficient single-pass version. The matching cost is the sum of absolute differences over small windows. Aggregation is performed by dynamic programming along paths in only 5 of 8 directions. Post filter as implemented in OpenCV. Dense results are created by hole-filling along scanlines.
SAD window: 3x3 pixel
Truncation value for pre-filter: 63
P1/P2: 8*3*3*3/32*3*3*3
Uniqueness ratio: 10
Speckle window size: 100
Speckle range: 32
Full DP: false
C/C++; 1 core, i7@3.3 GHz
07/28/14
6
SGBM1
H
2
43.4
157
55.1
168
29.3
179
28.1
158
33.0
133
81.1
223
34.2
136
57.0
148
52.3
154
18.6
113
65.6
179
59.5
162
41.2
151
41.9
136
43.6
134
64.8
172
Anonymous. Depth-based optimization for accurate stereo matching. ECCV 2022 submission 100.
We propose an accurate and lightweight convolutional neural network for stereo estimation with depth completion. The whole method consists of three parts. The first part consists of fully-convolutional densely connected layers that computes expressive features of rectified image pairs.
learning rate: 0.00006 for feature extraction and similarity and learning rate: 0.000006 for depth completion
Incorporating cues from top-down (holistic) scene understanding into existing bottom-up stereo reconstruction techniques (CoR - Chakrabarti et al. CVPR 2015).
Learned weightings (from 2006 dataset) for High Level Scene Cues. Default parameters for CoR. Images with max disp > 256 were downsampled before the SGM step of CoR.
Matlab and C++. single E5 core at 2.4GHz
04/24/16
31
HLSC_cor
H
2
44.7
160
42.9
143
22.0
149
34.8
179
36.4
143
53.0
129
44.5
184
60.8
168
61.1
187
26.2
160
53.0
153
59.2
161
42.7
160
50.0
176
54.3
169
44.0
109
Anonymous. A decomposition model for stereo matching. CVPR submission 2543.
OpenCV 2.4.8 StereoSGBM method, single-pass variant. Reimplementation and modification of H. Hirschmüller's SGM method (CVPR 2006; PAMI 2008).
OpenCV's "semi-global block matching" method; memory efficient single-pass version. The matching cost is the sum of absolute differences over small windows. Aggregation is performed by dynamic programming along paths in only 5 of 8 directions. Post filter as implemented in OpenCV. Dense results are created by hole-filling along scanlines.
SAD window: 3x3 pixel
Truncation value for pre-filter: 63
P1/P2: 8*3*3*3/32*3*3*3
Uniqueness ratio: 10
Speckle window size: 100
Speckle range: 32
Full DP: false
This approach is an adaptive local stereo-method. It is integrated into a hierarchical scheme, which exploits adaptive windows. Sub-pix disparities are estimated,but not refined.
L = 10
t = 35
medianK = [3 3]
censusK = [9 7]
lambda = 45;
Correlation with five, partly overlapping windows on Census transformed images using Hamming distance as matching cost. A left-right consistency check ensures unique matches and filtering small disparity segments removes outliers. Interpolation is done within image rows with the lowest, valid neighboring disparity.
Census window: 7x7 pixel
Correlation window: 9x9 pixel
LR-check: on
Min. segments: 200 pixel
Interpolation: horizontal, lowest neighbor
The images are Census transformed and the Hamming distance is used as pixelwise matching cost. Aggregation is performed by a kind of dynamic programming along 8 paths that go from all directions through the image. Small disparity patches are invalidated. Interpolation is also performed along 8 paths.
The method comprises two main steps. First, we use adaptive support weights for local matching. Apart from the color similarity and geometric distance, the adaptive weight distribution favors pixels in the block matching with smaller cost. Besides, we use a multiscale strategy with invalidation criteria to reduce match ambiguity and computational time.
Second, a global interpolation using a variational formulation is carried out. The energy functional penalizes deviations from the local disparity estimation at different scales.
Local approach (DAWA): 23x23 squared window, beta=11, lambda=6, gamma=4, pixel precision 1/4, three scales for multiscale procedure.
Variational model: alpha=1, gamma=5, phi1=30, phi2=15.
A deep-learning model PSMNU, modified based on PSMNet, produces initial disparity and uncertainty on the down-sampled image. SGBMP performs full resolution prediction based on the initial disparity and uncertainty.
PSMNU: max disparity 256, trained on Scene Flow dataset (Flyingthings3D & Monkaa) only, without data augmentation. SGBMP: \lambda_b = 3, \lambda_s = 0.1, \lambda_d = 0.1. For the initial prediction of PSMNU, images are down-sampled to 768x1024.
Apply FADNet as backbone, pre-train on SceneFlow for 90 epochs, finetuned on RVC 2020 samples (kitti2015, eth3d, middlebury)
lr=1e-4, finetuned for 1200 epochs, lr decayed by half every 200 epochs
Python; GTX 2070
06/02/21
146
FADNet_RVC
H
2
47.7
173
42.8
142
19.0
132
27.2
154
34.6
135
63.4
156
33.0
118
67.4
187
70.0
206
27.1
166
75.2
215
62.6
177
42.7
158
48.0
173
65.0
187
81.9
220
Anonymous. CCL-Stereo: Stereo matching via looking up coupled correlations. ICCSIP 2023 submission.
TITAN RTX
06/26/23
209
CCL-Stereo
F
3
48.0
174
69.4
211
27.5
171
26.9
151
51.7
193
92.7
237
9.91
5
71.0
196
55.2
167
26.9
163
58.0
158
60.5
167
39.6
140
45.7
157
75.1
223
61.4
160
Linghua Zeng and Xinmei Tian. CRAR: Accelerating stereo matching with cascaded regression and adaptive refinement. Submitted to Pattern Recognition, 2020.
Python;GTX1080Ti
02/20/20
109
CRAR
H
2
48.2
175
51.2
159
35.6
193
36.6
186
44.1
177
46.1
112
49.2
194
60.3
167
58.8
183
34.1
182
57.5
157
63.3
181
46.2
181
46.1
161
50.0
153
64.6
170
Anonymous. RANet++: Cost volume and correlation based network for efficient stereo matching. ICRA 2021 submission.
Tradeoff between performance and efficient with cost volume using 3D convolution and correlation with 2D convolution.
# of GPUs: 4,
Batch size per GPU: 4,
Learning rate with cosine decay: 1e-3
Optimizer: Adam
One Nvidia Tesla V100 GPU installed on a 32-core Intel CPU with 256GB memory.
06/04/21
147
RANet++
H
2
48.3
176
41.7
137
20.1
141
32.4
172
35.3
140
67.0
169
35.0
144
66.9
186
66.5
197
25.9
158
70.0
192
61.0
170
44.3
170
47.6
171
70.0
206
78.9
211
Lingyin Kong, Jiangping Zhu, and Sancong Ying. Local stereo matching using adaptive cross-region based guided image filtering with orthogonal weights. Submitted to Mathematical Problems in Engineering, 2020.
we propose an improved cost aggregation method, in which the matching cost volume is filtered by ACR-GIF-OW
A local matching technique utilizing SAD+Census cost measure and a recursive edge-aware aggregation through Successive Weighted Summation. Occlusion handling is provided via left-right cross check and a background favored filling.
smoothness parameter sigma = 24
5x5 Census window, Census weight=0.7, SAD weight=0.3, occlusion threshold=2
C/C++ one i5 core @3.10 GHz
04/09/15
17
PFS
F
3
48.7
178
76.8
235
40.4
202
25.3
145
61.5
210
81.5
226
47.7
190
66.4
184
49.4
146
19.1
120
48.6
142
60.5
169
42.4
157
54.1
193
41.3
127
53.8
136
Hao Liu, Hanlong Zhang, Xiaoxi Nie, Wei He, Dong Luo, Guohua Jiao and Wei Chen. Stereo matching algorithm based on two-phase adaptive optimization of AD-census and gradient fusion. IEEE RCAR 2021.
In this paper, an improved AD-Census algorithm is proposed to improve the matching ratio in some special regions. The proposed algorithm contains an optimization method and three similarity metrics.
A prior disparity image is calculated by matching a set of reliable support points and triangulating between them. A maximum a-posterior approach refines the disparities. The disparities for the left and right image are checked for consistency and disparity segments below a size of 50 pixels removed. (Improved results as of 9/14/2015 due to bug fix in color-to-gray conversion.)
Standard parameters of Libelas as provided with the MiddEval3-SDK.
C++/SSE; 1 core, i7@3.6 GHz
09/14/15
21
ELAS
F
3
50.1
180
69.4
211
27.5
171
26.9
151
51.7
193
92.7
237
36.1
151
71.0
196
55.2
167
26.9
163
58.0
158
60.5
167
39.6
140
45.7
157
75.1
223
61.4
160
Anonymous. RLStereo: Real-time stereo matching based on reinforcement learning. CVPR 2021 submission 4443.
Tensorflow 2.0; Nvidia GeForce Titan RTX GPU
11/12/20
133
RLStereo
H
2
50.9
181
45.2
149
27.4
170
38.2
192
48.6
186
79.8
219
56.1
220
57.9
152
54.4
162
39.9
195
78.5
224
58.4
156
43.3
161
48.6
175
70.1
207
40.8
97
OpenCV 2.4.8 StereoSGBM method, full variant (2 passes). Reimplementation of H. Hirschmüller's SGM method (CVPR 2006; PAMI 2008).
OpenCV's "semi-global block matching" method; memory-intensive 2-pass version, which can only handle the quarter-size images. The matching cost is the sum of absolute differences over small windows. Aggregation is performed by dynamic programming along paths in 8 directions. Post filter as implemented in OpenCV. Dense results are created by hole-filling along scanlines.
SAD window: 3x3 pixel
Truncation value for pre-filter: 63
P1/P2: 8*3*3*3/32*3*3*3
Uniqueness ratio: 10
Speckle window size: 100
Speckle range: 32
Full DP: true
A prior disparity image is calculated by matching a set of reliable support points and triangulating between them. A maximum a-posterior approach refines the disparities. The disparities for the left and right image are checked for consistency and disparity segments below a size of 50 pixels removed.
Updated ELAS submission as a baseline for the Robust Vision Challenge (http://robustvision.net), replacing the original ELAS (H) entry.
Standard parameters as provided with the MiddEval3-SDK and the Robust Vision Challenge stereo devkit.
We have collected 2000 pairs of stereo images with high accuracy disparity maps to fine-tune the network. Our goal is to improve the generalization performance of networks.
fine-tune num: 90000; the initial learning rate: 1e-3.
Our approach is an extension of the ELAS (from Geiger et al.) algorithm. We extract edges and sample our candidate support points along them. For every two consecutive valid support points we create a (straight) line segment. We force the triangulation to include the set of line segments (constrained Delaunay) for a better preservation of the disparity discontinuity at the edges.
Parameters as in the original ELAS algorithm.
For sampling candidate support points along the edge segments:
Adaptive sampling activated:
step = ceil(sqrt(img_diag)*0.5);
sampler(sqrt(step) / 2, step / 2, step / 2);
OpenCV 2.4.8 StereoSGBM method, single-pass variant. Reimplementation and modification of H. Hirschmüller's SGM method (CVPR 2006; PAMI 2008).
OpenCV's "semi-global block matching" method; memory efficient single-pass version. The matching cost is the sum of absolute differences over small windows. Aggregation is performed by dynamic programming along paths in only 5 of 8 directions. Post filter as implemented in OpenCV. Dense results are created by hole-filling along scanlines.
SAD window: 3x3 pixel
Truncation value for pre-filter: 63
P1/P2: 8*3*3*3/32*3*3*3
Uniqueness ratio: 10
Speckle window size: 100
Speckle range: 32
Full DP: false
C/C++; 1 core, i7@3.3 GHz
07/25/14
2
SGBM1
Q
1
53.6
187
56.3
172
43.0
209
43.4
199
41.4
167
76.0
207
41.0
174
63.4
174
61.8
189
38.7
194
74.7
214
64.3
186
48.1
187
53.5
191
61.2
181
78.0
204
Hector Vazquez, Madain Perez, Abiel Aguilar, Miguel Arias, Marco Palacios, Antonio Perez, Jose Camas, and Sabino Trujillo. Real-time multi-window stereo matching algorithm with fuzzy logic. Submitted to IET Computer Vision, 2020.
The propose a novel stereo matching algorithm with fuzzy logic and also implement it on a FPGA embedded system. We try to select the best window size of SAD for each pixel by leveraging fuzzy logic.
we propose a MST-based stereo
matching method using image edge and brightness
information due to the classical MST based methods were
used to produce the inaccurate matching weight in the
areas of image boundaries and similar color background.
We propose "DeepPruner", a real-time stereo matching algorithm, which combines the strength of deep network and search space pruning techniques. Towards this goal, we developed a differentiable PatchMatch module that allows us to discard most disparities and generates a sparse representation of the cost-volume. We then exploit this representation to learn which range to prune for each pixel. Our method achieves competitive results on KITTI / SceneFlow datasets while running in real-time at 62ms. Moreover, we obtain the first place (on overall rankings) on the Robust Vision Challenge. For more details, check out our paper and source code.
We proposed a robust disparity estimation network. Our major novelty compared to existing work is a novel usage of attention, which can handle scenes with different scenarios.
The RVC submission trained by quarter-resolution Middlebury + KITTI + ETH. After validation, we will go with quarter resolution instead of half-resolution
Lingyin Kong, Jiangping Zhu, and Sancong Ying. Stereo matching based on guidance image and adaptive support region. Submitted to Acta Optica Sinica, 2020.
We novelly formulate the scale transformation of cost volume as a Bayes inference and propose the inter-scale subnetwork to reliably and adaptively generate details under the guidance of geometric information.
we fine-tune the model pre-trained on Scene Flow for 300 epochs with the learning rate of 0.001 in the first 100 epochs and 0.0001 in the rest 1000 epochs.
No post processing (no filtering, no hole-filling, no interpolation) performed.
The concepts of intrinsic curves were revisited and used for:
- disparity search space reduction, resulting in 83% reduction of the disparity range (individually for each pixel) directly from the original resolution of the image without needing hierarchical search
- reducing the ambiguities due to occluded pixels by integrating occlusion clues explicitly into the global energy function as a soft prior
The final energy minimization was done using semi global approach along eight paths.
Matching (data) cost = census transform 7*9
Occlusion cost= from intrinsic curves curvature
Stereo matching process is attracted numbers of study in recent years. The process is unique and difficult due to visual discomfort occurred which contributed to effect of accuracy of disparity maps. By using multistage technique implemented most of Stereo Matching Algorithm; taxonomy by D. Scharstein and R. Szeliski, in this paper proposed new improvement algorithm of stereo matching by using the effect of Adaptive Weighted Bilateral Filter as main filter in cost aggregation stage which able contribute edge-preserving factor and robust against plain colour region. With some improvement parameters in matching cost computation stage where windows size of sum of absolute different (SAD) and thresholds adjustment was applied and Median Filter as main filter in refinement disparity map’s stage may overcome the limitation of disparity map accuracy. Evaluation on indoor datasets, latest (2014) Middlebury dataset were used to prove that Adaptive Weighted Bilateral Filter effect applied on proposed algorithm resulted smooth disparity maps and achieved good processing time.
Numerous CNN algorithms focus on the pixel-wise matching cost computation, which is the important building block for many state-of-the-art algorithms. However, these architectures are limited to small and single scale receptive fields and use traditional methods for cost aggregation or even ignore cost aggregation. In this paper, we propose a novel architecture called cascaded multi-scale and multi-dimension network (MSMD) to take them both into consideration. Firstly, we propose a new multi-scale matching cost computation sub-network, in which two different sizes of receptive fields are implemented parallelly. In this way, the network can make the best use of both variants to balance the trade-off between the increase of receptive field and the loss of details. Furthermore, we show that our multi-dimension aggregation sub-network which contains 2D convolution and 3D convolution operations can provide rich context and semantic information for estimating an accurate initial disparity.
Stereo matching algorithm based on multi-cost computation with hybrid aggregation using random walk and image segmentation with filtering in refinement stage.
Windows 10; Xeon(R) 8core @2.6GHz; 64G RAM; Matlab; C/C++;
08/29/22
175
MCP-HA-VQ
Q
1
64.1
214
73.0
231
55.3
219
60.0
216
62.8
215
70.1
178
52.4
203
70.7
195
66.9
198
58.1
215
68.6
186
70.6
196
64.4
215
64.9
210
67.4
194
74.7
194
Hao Li, Yanwei Sun, and Li Sun. Edge-preserved disparity estimation with piecewise cost aggregation. Submitted to the International Journal of Geo-Information, 2019.
The cost aggregated paths are divided by edge pixels in the edge disparity map, and cost aggregation is calculated independently in each sub-path.
Stereo matching algorithm based on edge-preserving filter at cost aggregation and image segmentation at disparity refinement stage.
iterative n =3, k segmentation =250
C++
01/15/17
44
IGF
Q
1
66.2
216
69.9
217
57.1
225
61.3
220
65.8
221
72.0
194
55.4
214
72.7
205
69.4
204
60.4
220
73.2
203
75.3
216
67.0
221
67.1
221
69.0
199
74.5
192
Hong Li and Chunbo Cheng. Adaptive weighted matching cost based on sparse representation. Submitted to IEEE TIP, 2018.
This paper proposes a novel non-data-driven matching cost for dense correspondence in view of sparse representation. This new matching cost can separate the source of impact such as illuminations and exposures, thus making it more suitable and selective for stereo matching. In addition, the new matching cost can be used as a adaptive weight in the process of cost calculation, and can improve the accuracy of the matching costs by weighting.
In stereo matching cost filtering methods and energy minimization algorithms are considered as two different techniques. Due to their global extend energy minimization methods obtain good stereo matching results. However, they tend to fail in occluded regions, in which cost filtering approaches obtain better results. In this paper we intend to combine both approaches with the aim to improve overall stereo matching results.
We propose to perform stereo matching as a two-step energy minimization algorithm. We consider two MRF models: a fully connected model defined on the complete set of pixels in an image and a conventional locally connected model. We solve the energy minimization problem for the fully connected model, after which the marginal function of the solution is used as the unary potential in the locally connected MRF model.
Only gradient component (6D vector) of color images is used
C++, Intel® Core™ i5-4300U 1.9-GHz CPU
01/21/15
15
TSGO
F
3
66.6
218
63.2
196
48.1
213
51.4
208
73.5
239
78.8
216
39.0
163
78.8
229
76.0
226
66.0
239
82.7
231
75.2
215
68.0
228
68.8
230
74.7
222
77.1
203
Madiha Zahari. A new cost volume estimation using modified CT. Submitted to the Bulletin of Electrical Engineering and Informatics (BEEI), paper ID 4122, 2022.
Visual Studio c++, Intel(R) Core(TM) i7-8565U CPU @ 1.80GHz 1.99 GHz
Introduce an effective GPU-based calculation method for ZNCC.
NCC: 5x5
Jetson Tx2
04/11/22
166
Z2ZNCC
Q
1
67.1
221
69.7
215
58.6
231
64.8
235
65.9
222
71.3
189
55.4
215
72.8
206
70.6
209
63.2
233
73.0
202
75.1
213
68.4
230
68.5
228
69.9
204
71.8
187
Yuli Fu, Kaimin Lai, Weixiang Chen, and Youjun Xiang. A pixel pair based encoding pattern for stereo matching via an adaptively weighted cost. Submitted to IET Image Processing, 2020.
A novel encoding pattern, which is designed for the situation of radiometric distortion, is proposed. The pattern is applied for stereo matching cost function.
We propose an approach for real-time embedded stereo processing on ARM and CUDA-enabled devices, which is based on the popular and widely used Semi-Global Matching algorithm. In this, we propose an optimization of the algorithm for embedded CUDA GPUs, by using massively parallel computing, as well as using the NEON intrinsics to optimize the algorithm for vectorized SIMD processing on embedded ARM CPUs.
This method utilizes a deep learning technique to perform a self-guided cost aggregation which does not require any guidance color image.
Matlab; i7-4770 @ 3.40 GHz; GTX 1080 GPU
03/23/17
49
DSGCA
Q
1
67.5
225
71.9
225
59.7
234
61.9
223
66.5
227
72.9
197
56.1
219
73.8
208
70.9
210
61.6
227
74.2
212
75.6
218
67.4
223
69.6
233
71.0
213
74.8
195
Rafael Brandt, Nicola Strisciuglio, and Nicolai Petkov. MTStereo 2.0: Improved accuracy of stereo depth estimation. ICPR 2020 submission.
The method is based on a Max-tree hierarchical representation of image pairs, which we use to identify matching regions along image scan-lines.
The number of color quantization levels was set to 16. α was set to 0.8. The minimum (or maximum) width of nodes to be matched was set to 0 (or 1/2 of the input image width). Matched node levels S was set to {1, 0}. The maximum neighborhood size ω_γ was set to 10. The size of the Gaussian kernel used to aggregate the cost volume was 21. The minimum confidence percentage parameter ω_Π was set to 12. In guided pixel refinement, ω_ω was set to 12% when sparse disparity maps were generated.
i7 8565U (4 cores)
01/05/20
106
MTS2
F
3
67.6
226
70.7
219
36.5
195
54.4
211
69.3
234
98.8
242
58.2
227
81.3
237
77.4
234
48.1
207
91.7
239
82.7
238
62.0
210
58.9
196
81.9
236
91.5
239
Xiaowei Yang. A light-weight stereo matching network based on multi-scale features fusion and robust disparity refinement. Submitted to IET Image Processing, 2022.
In recent years, convolutional-neural-network based stereo matching methods have achieved significant
gains compared to conventional methods in terms of both speed and accuracy. Current state-of-the-art disparity
estimation algorithms require many parameters and large amounts of computational resources and are not suited to
applications on edge devices. In this paper, we propose an end-to-end light-weight network (LWNet) for fast stereo
matching, which consists of an efficient backbone with multi-scale feature fusion for feature extraction, a 3D U-Net
aggregation architecture for disparity computation and a color guidance in 2D CNN for disparity refinement.
This article presents a disparity map algorithm to improve the depth map estimation based on Census Transform and hierarchical segment-tree on each block.The stereo matching algorithm presented in this study comprises of four steps: Cost Computation, Cost
Aggregation, Optimization, and Post-Processing, all of which will refine the final disparity map.
CostAlpha = 0.3;
CEN-WND = 9x11;
k = 1600;
LR checking = Yes
PY_LVL = 3.
C++, a personal PC with a CPU i7 8700@3.2 GHz, an RTX 2070 SUPER, and 16GB RAM.
This is a segmentation based stereo matching algorithm using an adaptive multi-cost approach, which is exploited for obtaining accuracy disparity maps.
Chenglong Xu, Chengdong Wu, Daokui Qu, Haibo Sun and Jilai Song. Accurate and efficient stereo matching by log-angle and pyramid-tree. Submitted to IEEE TCSVT, 2020.
Combined bearings-only cost metric and Cross-regional connection based aggregation.
C++;Intel Core i7-4558U@2.8 GHz
09/03/20
128
LPSM
Q
1
68.7
231
68.2
206
56.9
224
62.9
229
66.5
228
66.7
168
51.5
201
79.5
232
77.0
231
61.8
228
68.5
185
85.2
239
69.8
236
69.3
232
74.1
219
89.9
237
Anonymous. STTR. RVC 2020 submission.
GPU
08/14/20
126
STTRV1_RVC
Q
1
69.1
232
57.5
176
59.4
233
64.2
231
61.8
212
73.2
199
62.9
237
69.7
194
69.9
205
71.9
242
86.8
234
76.1
220
62.1
213
73.3
238
82.2
238
78.1
206
Yun Xie, Shaowu Zheng, and Weihua Li. Feature-guided spatial attention upsampling for real-time stereo matching. Submitted to IEEE MultiMedia, 2020.
Our method is local matching approach using the Guided Filter for cost aggregation. We give appropriate the Guided Filter size for each pixel in input image by the Filter Size Map computed by using the DoG Kernel.
Parameters for Filter Size Map computation:
DoGparam.scalesize = 25 (index of scale space)
DoGparam.mfsize = 1 (window size for Filter Size Map optimization)
Parameters for Guided Filter:
eps = 0.001
Parameters for cost computation:
gamma = 0.11 (Weight of cost)
Parameters for Bilateral Filter in disparity map optimization:
gamma_c = 1
gamma_d = 11
r_median = 19
Matlab, core-i5 @3.0GHz (2 cores, 4 threads)
06/14/17
52
DoGGuided
Q
1
70.7
236
72.9
229
60.4
238
64.5
233
68.2
232
76.3
209
56.0
217
81.5
238
78.8
235
63.7
236
79.7
227
80.2
234
69.7
234
70.0
234
73.5
218
87.0
233
Kang Zhang, Jiyang Li, Yijing Li, Weidong Hu, Lifeng Sun, and Shiqiang Yang. Binary stereo matching. ICPR 2012.
no post processing is used
the same with the original paper.
C/C++ single thread Intel(R) Core(TM)2 Duo CPU P7370 @ 2.00GHz
The computation of the sparse disparity maps is achieved by means of a 3D diffusion of the costs contained in the disparity space volume. The watershed segmentations of the left and right views control the diffusion process and valid measurements are obtained by cross-checking.
The estimation of the dense disparity maps uses the sparse measurements as control points and is driven by a 3D watershed separating the disparity space volume into foreground and background pixels.
The algorithm is based on a hierarchical representation of image pairs which is used to restrict disparity search range. We propose a cost function that takes into account region contextual information and a cost aggregation method that preserves disparity borders.
An energy minimization framework for disparity estimation where energy function consists of intensity matching cost, feature matching cost, IGMRF prior and sparsity priors.
Manually set
MATLAB 2014 @2.22 Ghz
10/23/16
39
SIGMRF
Q
1
81.9
243
80.6
241
65.8
242
83.7
243
82.2
243
99.7
245
63.5
238
93.2
243
91.6
243
68.8
241
97.7
243
96.4
243
79.1
242
83.3
243
78.4
230
94.7
241
Average disparity over all training images of the ROB 2018 stereo challenge.
This submission is a baseline for the Robust Vision Challenge ROB 2018. Each pixel is set to the average disparity of the pixels at the same location in the training images. No test image information is used.
03/23/18
64
AVERAGE_ROB
H
2
98.8
244
98.0
245
98.1
245
98.3
244
99.0
244
99.0
243
98.8
244
98.9
244
99.1
245
99.4
244
99.4
244
99.5
244
99.8
244
96.6
244
98.8
245
99.4
245
Median disparity over all training images of the ROB 2018 stereo challenge.
This submission is a baseline for the Robust Vision Challenge ROB 2018. Each pixel is set to the median disparity of the pixels at the same location in the training images. No test image information is used.
