Pith. sign in

REVIEW 3 major objections 43 references

LiteMatch achieves competitive zero-shot stereo matching across domains with a lightweight model by stabilizing cost volumes instead of using heavy 3D convolutions.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-07-01 06:11 UTC pith:LFN7DWGF

load-bearing objection LiteMatch proposes a lightweight stereo setup with CVCE, HFE, and CVC-Loss for zero-shot gains, but the loss's claimed stabilization effect has no visible supporting analysis or ablations. the 3 major comments →

arxiv 2606.31636 v1 pith:LFN7DWGF submitted 2026-06-30 cs.CV

LiteMatch: Lightweight Zero-Shot Stereo Matching via Cost Volume Stabilization

Md Raqib Khan , Santosh Kumar Vipparthi , Subrahmanyam Murala This is my paper
classification cs.CV
keywords stereo matchingzero-shot generalizationcost volumelightweight modelCVC-Lossdisparity estimationdomain generalizationfrequency cues
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces LiteMatch as a framework that combines a cross-view correspondence encoder and a high-frequency encoder to build cost volumes, then applies a dedicated consistency loss to sharpen the disparity probability distributions. This design keeps the total parameter count between 3.36 million and 9.58 million while delivering end-point error and D1 scores that remain competitive on Scene Flow, KITTI, Middlebury, ETH3D, and DrivingStereo without any target-domain training. The central argument is that explicit stabilization of the cost volume removes the need for expensive 3D processing and enables the model to generalize where heavier architectures typically fail. A sympathetic reader would care because practical stereo systems often cannot afford large models or repeated fine-tuning when the camera setup or scene type changes.

Core claim

LiteMatch shows that a lightweight stereo matching pipeline reaches strong zero-shot performance on five standard benchmarks by replacing 3D cost-volume convolutions with two complementary encoders plus the Cost Volume Consistency Loss, which is a voxel-wise binary cross-entropy term applied after softmax normalization to drive disparity probabilities toward sharp, unimodal peaks.

What carries the argument

The Cost Volume Consistency Loss (CVC-Loss), a voxel-wise binary cross-entropy objective on softmax-normalized cost distributions that encourages stable, unimodal disparity probabilities and thereby supports rapid convergence and cross-domain generalization.

Load-bearing premise

The binary cross-entropy loss on softmax cost distributions will reliably produce stable unimodal disparity probabilities that enable zero-shot generalization without 3D convolutions.

What would settle it

A clear drop below the reported EPE and D1 scores on one or more of the five test datasets when the CVC-Loss is removed or replaced while keeping all other components fixed.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Stereo matching becomes feasible on devices with tight memory and compute budgets.
  • Zero-shot deployment across camera rigs and environments no longer requires heavy recurrent refinement or domain adaptation.
  • A simple lightweight refinement stage is sufficient to reach full-resolution output once the cost volume is stabilized.
  • Parameter counts as low as 3.36 million remain compatible with competitive accuracy on multiple public benchmarks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same stabilization approach could be tested on related dense matching problems such as optical flow where cost volumes are also central.
  • Removing the dependence on 3D convolutions may reduce power draw enough to enable longer battery life in mobile vision applications.
  • Explicit regularization of probability sharpness might substitute for architectural complexity in other correspondence tasks.
  • A follow-up experiment could measure whether the same loss improves generalization when the training and test domains differ more extremely than the five datasets used here.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 0 minor

Summary. The paper introduces LiteMatch, a lightweight stereo matching network (3.36M–9.58M parameters) that uses a Cross-View Correspondence Encoder (CVCE), a High-Frequency Encoder (HFE) based on FFT, a voxel-wise binary cross-entropy Cost Volume Consistency Loss (CVC-Loss) applied to softmax-normalized cost volumes, and a lightweight refinement module. It claims that CVC-Loss produces sharp unimodal disparity probabilities that stabilize the cost volume, enabling competitive zero-shot EPE and D1 performance on Scene Flow, KITTI, Middlebury, ETH3D, and DrivingStereo without 3D convolutions or heavy recurrent refinement.

Significance. If the central claim holds, the work would demonstrate that cost-volume stabilization via a simple BCE loss can deliver domain-general stereo matching in a parameter-efficient regime, reducing reliance on 3D convolutions and thereby lowering memory and compute demands for practical deployment.

major comments (3)
  1. [Abstract] Abstract: the assertion that CVC-Loss 'encourages sharp and unimodal disparity probabilities' and thereby 'promotes stable cost distributions' is presented without any supporting quantitative evidence (entropy statistics, modality counts, peak-sharpness metrics, or distribution visualizations).
  2. [Abstract] Abstract: competitive EPE and D1 scores are claimed across five datasets, yet the text provides no baselines, error bars, dataset splits, or ablation isolating CVC-Loss from CVCE/HFE; this prevents verification of the zero-shot generalization result.
  3. [Abstract] Abstract: the central claim that the method 'achieves strong zero-shot generalization through cost volume stabilization' rests on the unverified assumption that voxel-wise BCE on softmax costs yields unimodal distributions; without ablations or distribution analysis this link remains unestablished.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for highlighting these issues in the abstract. We will make revisions to provide the necessary supporting evidence and details as outlined in our responses below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the assertion that CVC-Loss 'encourages sharp and unimodal disparity probabilities' and thereby 'promotes stable cost distributions' is presented without any supporting quantitative evidence (entropy statistics, modality counts, peak-sharpness metrics, or distribution visualizations).

    Authors: We agree with the referee that the abstract currently lacks supporting quantitative evidence for the effect of CVC-Loss. We will revise the manuscript by adding entropy statistics, modality counts, and distribution visualizations to support this claim, and update the abstract to reference these new analyses. revision: yes

  2. Referee: [Abstract] Abstract: competitive EPE and D1 scores are claimed across five datasets, yet the text provides no baselines, error bars, dataset splits, or ablation isolating CVC-Loss from CVCE/HFE; this prevents verification of the zero-shot generalization result.

    Authors: We agree that the abstract should include more details for verification. We will revise the abstract to include specific EPE and D1 scores with baseline comparisons, mention the dataset splits used, and note the ablations isolating CVC-Loss. Error bars can be included if the referee deems necessary. revision: yes

  3. Referee: [Abstract] Abstract: the central claim that the method 'achieves strong zero-shot generalization through cost volume stabilization' rests on the unverified assumption that voxel-wise BCE on softmax costs yields unimodal distributions; without ablations or distribution analysis this link remains unestablished.

    Authors: We acknowledge that the assumption needs verification through analysis. We will add ablations and distribution analysis to the manuscript to establish the link between the BCE loss and unimodal distributions, and revise the abstract accordingly. revision: yes

Circularity Check

0 steps flagged

No circularity detected; derivation is self-contained empirical method

full rationale

The paper introduces CVCE, HFE encoders and CVC-Loss (voxel-wise BCE on softmax-normalized costs) as architectural choices motivated by standard practices for cross-view correspondence and distribution sharpening. Generalization results are reported as empirical outcomes on held-out datasets (Scene Flow, KITTI, etc.) rather than derived by construction from the loss definition. No equation reduces a claimed prediction to a fitted parameter, no uniqueness theorem is invoked via self-citation, and the loss objective does not tautologically guarantee the zero-shot transfer performance. The chain remains independent of its inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

Only abstract available, so ledger is limited to explicitly mentioned components; CVC-Loss is treated as an invented training objective with no independent evidence supplied.

axioms (2)
  • standard math Softmax produces valid probability distributions over disparities
    Invoked when applying binary cross-entropy to cost volume outputs
  • domain assumption Encouraging unimodal sharp distributions improves generalization
    Central premise of CVC-Loss design
invented entities (1)
  • CVC-Loss no independent evidence
    purpose: Stabilize cost volume by penalizing non-sharp disparity distributions
    New loss function introduced to replace heavy 3D convolutions

pith-pipeline@v0.9.1-grok · 5814 in / 1356 out tokens · 44087 ms · 2026-07-01T06:11:37.643681+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of LiteMatch: Lightweight Zero-Shot Stereo Matching via Cost Volume Stabilization." pith.science (2026). https://pith.science/paper/LFN7DWGF

@misc{pith2026260631636,
  author       = {Pith},
  title        = {Pith review of: LiteMatch: Lightweight Zero-Shot Stereo Matching via Cost Volume Stabilization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LFN7DWGF}},
  note         = {Machine review of arXiv:2606.31636}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Despite rapid progress in learning-based stereo matching, high accuracy is often achieved at the cost of heavy backbones and computationally intensive 3D cost volume processing, resulting in substantial memory and runtime overhead. More critically, these methods frequently struggle to generalize across domains, limiting their practical deployment. We present \textit{LiteMatch}, a lightweight stereo matching framework that achieves strong zero-shot generalization through cost volume stabilization-without expensive 3D convolutions. LiteMatch employs two complementary encoders: a Cross-View Correspondence Encoder (CVCE) to capture global cross-view interactions, and a High-Frequency Encoder (HFE) that enhances fine structural details via FFT-based frequency cues. To stabilize the cost volume, we introduce the \textit{Cost Volume Consistency Loss (CVC-Loss)}, a voxel-wise binary cross-entropy objective applied to softmax-normalized cost distributions. By encouraging sharp and unimodal disparity probabilities, CVC-Loss promotes stable cost distributions and enables rapid convergence. A lightweight refinement module further produces sharp full-resolution disparities with low-iteration updates, avoiding heavy recurrent refinement. With a flexible design ranging from 3.36M to 9.58M parameters, LiteMatch achieves exceptional zero-shot generalization, delivering competitive EPE and D1 performance across Scene Flow, KITTI, Middlebury, ETH3D, and DrivingStereo. Our results establish that lightweight architectures can indeed generalize across domains without sacrificing accuracy. \href{https://mdraqibkhan.github.io/Litematch}{\textcolor{blue}{Code}}

Figures

Figures reproduced from arXiv: 2606.31636 by Md Raqib Khan, Santosh Kumar Vipparthi, Subrahmanyam Murala.

Figure 1
Figure 1. Figure 1: Zero-shot cross-domain generalization. (a) & (c) LiteMatch produces cleaner disparity maps than state-of-the-art methods on challenging real-world scenes, (b) de￾spite having significantly fewer parameters. This demonstrates superior efficiency and robustness to domain shift. 3D convolutions. This limitation has motivated the exploration of more efficient paradigms. RAFT-Stereo [17] pioneered a different s… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of LiteMatch. Our framework employs a progressive two-stage scheme: (1) coarse training jointly optimizes the feature extractor and cost volume using our CVC-Loss to establish stable cost volume representations; (2) refinement training freezes Stage 1 components and trains only the transformer-based refinement module with L1 loss to recover fine details. The default model performs efficient single… view at source ↗
Figure 3
Figure 3. Figure 3: Cost volume distributions and entropy for a selected pixel. CE (a) shows higher entropy (2.74) than CVC (b, 1.77), with the target one-hot (c) at 0.0. Training dy￾namics: (d) CVC reduces uncertainty faster, and (e) maintains higher confidence via peak-to-second-peak ratios. CVC-Loss Formulation: Given cost volume C ∈ R H×W×D and ground-truth Dgt, we obtain probabilities Pi,j,d = Softmaxd(Ci,j,d). Let d ∗ (… view at source ↗
Figure 4
Figure 4. Figure 4: Iterative Refinement Comparison. Left: Disparity maps at iterations 4, 16, and 32 for MonSter, Selective-IGEV, and LiteMatch, showing convergence and structural clarity. Right: KITTI-2015 D1 error vs. refinement iterations, demonstrating LiteM￾atch’s early stabilization and efficient disparity refinement [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative comparison of LiteMatch against state-of-the-art stereo methods, illustrating sharper disparity estimation and cross-domain robustness. Rows 1–2 show results on ETH3D leaderboard samples (fine-tuned), rows 3–4 on KITTI. The final two rows present outputs from the base models and single-iteration refinements of the respective methods for reference. Comparison with Foundation Models. Recent ETH3D… view at source ↗
Figure 6
Figure 6. Figure 6: Visual ablation of CVC-Loss. From left to right: left input image; raw disparity without CVC-Loss (noisy, ambiguous); raw disparity with CVC-Loss (clean, sharp, low-entropy); final stage-1 refined cost volume disparity (clean and preserved details) [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Top row: Complementary feature activations from the Dual Encoder(E1)captures smooth, globally coherent regions, while E2 focuses on sparse, high-frequency edge and texture details; their fusion provides a balanced representation for disparity estimation. Bottom row: Corresponding initial disparity maps raw correlation without CVC-Loss(a) is noisy, without HFE (E2)(b) is noisy, while fused features and CVC-… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [1]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Badki, A., Troccoli, A., Kim, K., Kautz, J., Sen, P., Gallo, O.: Bi3d: Stereo depth estimation via binary classifications. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 1600–1608 (2020) 4

    work page 2020

  2. [2]

    Science China Information Sciences63(11), 212101 (2020) 16

    Bao, W., Wang, W., Xu, Y., Guo, Y., Hong, S., Zhang, X.: Instereo2k: a large real dataset for stereo matching in indoor scenes. Science China Information Sciences63(11), 212101 (2020) 16

    work page 2020

  3. [3]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Bartolomei, L., Tosi, F., Poggi, M., Mattoccia, S.: Stereo anywhere: Robust zero-shot deep stereo matching even where either stereo or mono fail. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 1013–1027 (2025) 4

    work page 2025

  4. [4]

    SIAM review60(2), 223–311 (2018) 18, 19

    Bottou, L., Curtis, F.E., Nocedal, J.: Optimization methods for large-scale machine learning. SIAM review60(2), 223–311 (2018) 18, 19

    work page 2018

  5. [5]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Chang, J.R., Chen, Y.S.: Pyramid stereo matching network. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 5410–5418 (2018) 1, 10

    work page 2018

  6. [6]

    In: Proceedings oftheIEEE/CVFConferenceonComputerVisionandPatternRecognition

    Chen, Z., Long, W., Yao, H., Zhang, Y., Wang, B., Qin, Y., Wu, J.: Mocha- stereo: Motif channel attention network for stereo matching. In: Proceedings oftheIEEE/CVFConferenceonComputerVisionandPatternRecognition. pp. 27768–27777 (2024) 4, 10, 11

    work page 2024

  7. [7]

    The Visual Computer 41(1), 409–421 (2025) 1

    Chen, Z., Zhao, Y., He, J., Lu, Y., Cui, Z., Li, W., Zhang, Y.: Feature distri- bution normalization network for multi-view stereo. The Visual Computer 41(1), 409–421 (2025) 1

    work page 2025

  8. [8]

    Monster++: Unified stereo matching, multi-view stereo, and real-time stereo with monodepth priors,

    Cheng, J., Liu, L., Xu, G., Wang, X., Zhang, Z., Deng, Y., Zang, J., Chen, Y., Cai, Z., Yang, X.: Monster: Marry monodepth to stereo unleashes power. arXiv preprint arXiv:2501.08643 (2025) 2, 4, 10, 11, 17

    work page arXiv 2025

  9. [9]

    arXiv preprint arXiv:2007.04532 (2020) 18, 19

    Faghri, F., Duvenaud, D., Fleet, D.J., Ba, J.: A study of gradient variance in deep learning. arXiv preprint arXiv:2007.04532 (2020) 18, 19

    work page arXiv 2007

  10. [10]

    In: 2012 IEEE conference on computer vision and pattern recognition

    Geiger, A., Lenz, P., Urtasun, R.: Are we ready for autonomous driving? the kitti vision benchmark suite. In: 2012 IEEE conference on computer vision and pattern recognition. pp. 3354–3361. IEEE (2012) 11

    work page 2012

  11. [11]

    Guo, X., Yang, K., Yang, W., Wang, X., Li, H.: Group-wise correlation stereonetwork.In:ProceedingsoftheIEEEConferenceonComputerVision and Pattern Recognition. pp. 3273–3282 (2019) 10

    work page 2019

  12. [12]

    Jiang, H., Lou, Z., Ding, L., Xu, R., Tan, M., Jiang, W., Huang, R.: Defom- stereo: Depth foundation model based stereo matching (2025) 2, 10, 11, 17

    work page 2025

  13. [13]

    In: 2023 International Joint Conference on Neural Networks (IJCNN)

    Khan, M.R., Kulkarni, A., Phutke, S.S., Murala, S.: Underwater image en- hancement with phase transfer and attention. In: 2023 International Joint Conference on Neural Networks (IJCNN). pp. 1–8. IEEE (2023) 5

    work page 2023

  14. [14]

    In: 2025 IEEE/CVF Winter Conference on Appli- cations of Computer Vision (WACV)

    Khan, M.R., Negi, A., Kulkarni, A., Phutke, S.S., Vipparthi, S.K., Murala, S.: Phaseformer: Phase-based attention mechanism for underwater image LiteMatch Supplementary Material 21 restoration and beyond. In: 2025 IEEE/CVF Winter Conference on Appli- cations of Computer Vision (WACV). pp. 9618–9629. IEEE (2025) 5

    work page 2025

  15. [15]

    In: Proceedings of the IEEE/CVF winter conference on applications of computer vision

    Khan, R., Mishra, P., Mehta, N., Phutke, S.S., Vipparthi, S.K., Nandi, S., Murala, S.: Spectroformer: Multi-domain query cascaded transformer net- work for underwater image enhancement. In: Proceedings of the IEEE/CVF winter conference on applications of computer vision. pp. 1454–1463 (2024) 5

    work page 2024

  16. [16]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Li, J., Wang, P., Xiong, P., Cai, T., Yan, Z., Yang, L., Liu, J., Fan, H., Liu, S.: Practical stereo matching via cascaded recurrent network with adap- tive correlation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 16263–16272 (2022) 16, 18, 19

    work page 2022

  17. [17]

    In: 2021 International Conference on 3D Vision (3DV)

    Lipson, L., Teed, Z., Deng, J.: Raft-stereo: Multilevel recurrent field trans- forms for stereo matching. In: 2021 International Conference on 3D Vision (3DV). pp. 218–227. IEEE (2021) 2, 4, 8, 10, 17

    work page 2021

  18. [18]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Mayer, N., Ilg, E., Hausser, P., Fischer, P., Cremers, D., Dosovitskiy, A., Brox, T.: A large dataset to train convolutional networks for disparity, opti- cal flow, and scene flow estimation. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4040–4048 (2016) 10

    work page 2016

  19. [19]

    In: Pro- ceedingsoftheIEEEconferenceoncomputervisionandpatternrecognition

    Menze, M., Geiger, A.: Object scene flow for autonomous vehicles. In: Pro- ceedingsoftheIEEEconferenceoncomputervisionandpatternrecognition. pp. 3061–3070 (2015) 11

    work page 2015

  20. [20]

    In: Proceedings of the IEEE international conference on computer vision workshops

    Pang,J.,Sun,W.,Ren,J.S.,Yang,C.,Yan,Q.:Cascaderesiduallearning:A two-stage convolutional neural network for stereo matching. In: Proceedings of the IEEE international conference on computer vision workshops. pp. 887–895 (2017) 19

    work page 2017

  21. [21]

    In: Advances in Neural Information Processing Systems (NeurIPS) (2019) 9

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al.: Pytorch: An imper- ative style, high-performance deep learning library. In: Advances in Neural Information Processing Systems (NeurIPS) (2019) 9

    work page 2019

  22. [22]

    In: Pattern Recognition: 36th German Conference, GCPR 2014, Münster, Germany, September 2-5, 2014, Proceedings 36

    Scharstein, D., Hirschmüller, H., Kitajima, Y., Krathwohl, G., Nešić, N., Wang, X., Westling, P.: High-resolution stereo datasets with subpixel- accurate ground truth. In: Pattern Recognition: 36th German Conference, GCPR 2014, Münster, Germany, September 2-5, 2014, Proceedings 36. pp. 31–42. Springer (2014) 11

    work page 2014

  23. [23]

    International journal of computer vision 47(1), 7–42 (2002) 1

    Scharstein, D., Szeliski, R.: A taxonomy and evaluation of dense two-frame stereo correspondence algorithms. International journal of computer vision 47(1), 7–42 (2002) 1

    work page 2002

  24. [24]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Schops, T., Schonberger, J.L., Galliani, S., Sattler, T., Schindler, K., Polle- feys, M., Geiger, A.: A multi-view stereo benchmark with high-resolution images and multi-camera videos. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 3260–3269 (2017) 11

    work page 2017

  25. [25]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Shen, Z., Dai, Y., Rao, Z.: Cfnet: Cascade and fused cost volume for robust stereo matching. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 13906–13915 (2021) 4

    work page 2021

  26. [26]

    learning

    Tan, C., Zhao, Y., Wei, S., Gu, G., Liu, P., Wei, Y.: Frequency-aware deep- fake detection: Improving generalizability through frequency space domain 22 Khan et al. learning. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 38, pp. 5052–5060 (2024) 6

    work page 2024

  27. [27]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Tankovich, V., Hane, C., Zhang, Y., Kowdle, A., Fanello, S., Bouaziz, S.: Hitnet: Hierarchical iterative tile refinement network for real-time stereo matching. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 14362–14372 (2021) 4

    work page 2021

  28. [28]

    In: European conference on computer vision

    Teed, Z., Deng, J.: Raft: Recurrent all-pairs field transforms for optical flow. In: European conference on computer vision. pp. 402–419. Springer (2020) 8

    work page 2020

  29. [29]

    In: Proceedings of the 30th ACM international conference on multimedia

    Wang, J., Du, R., Chang, D., Liang, K., Ma, Z.: Domain generalization via frequency-domain-based feature disentanglement and interaction. In: Proceedings of the 30th ACM international conference on multimedia. pp. 4821–4829 (2022) 5

    work page 2022

  30. [30]

    Wang, W., Zhu, D., Wang, X., Hu, Y., Qiu, Y., Wang, C., Hu, Y., Kapoor, A., Scherer, S.: Tartanair: A dataset to push the limits of visual slam (2020) 16

    work page 2020

  31. [31]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Wang, X., Xu, G., Jia, H., Yang, X.: Selective-stereo: Adaptive frequency information selection for stereo matching. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 19701–19710 (2024) 2, 4, 5, 10, 11, 17

    work page 2024

  32. [32]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Wang, Y., Wang, L., Zhang, C., Zhang, Y., Zhang, Z., Ma, A., Fan, C., Lam, T.L., Hu, J.: Learning robust stereo matching in the wild with se- lective mixture-of-experts. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 21276–21287 (2025) 10, 11, 17

    work page 2025

  33. [33]

    org/abs/1802.0379619

    Weinshall, D., Cohen, G., Amir, D.: Curriculum learning by transfer learn- ing: Theory and experiments with deep networks (2018),https://arxiv. org/abs/1802.0379619

    work page arXiv 2018

  34. [34]

    Foundationstereo: Zero-shot stereo matching,

    Wen, B., Trepte, M., Aribido, J., Kautz, J., Gallo, O., Birchfield, S.: Foun- dationstereo: Zero-shot stereo matching. arXiv preprint arXiv:2501.09898 (2025) 2, 4, 11, 17

    work page arXiv 2025

  35. [35]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Xu, G., Wang, X., Ding, X., Yang, X.: Iterative geometry encoding vol- ume for stereo matching. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 21919–21928 (2023) 2, 10, 11

    work page 2023

  36. [36]

    In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition

    Xu, P., Xiang, Z., Qiao, C., Fu, J., Pu, T.: Adaptive multi-modal cross- entropy loss for stereo matching. In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition. pp. 5135–5144 (2024) 4

    work page 2024

  37. [37]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Yang, G., Song, X., Huang, C., Deng, Z., Shi, J., Zhou, B.: Drivingstereo: A large-scale dataset for stereo matching in autonomous driving scenarios. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 899–908 (2019) 11

    work page 2019

  38. [38]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Yao, C., Yu, L., Liu, Z., Zeng, J., Wu, Y., Jia, Y.: Diving into the fusion of monocular priors for generalized stereo matching. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 14887–14897 (2025) 4 LiteMatch Supplementary Material 23

    work page 2025

  39. [39]

    In: CVPR (2022) 6

    Zamir, S.W., Arora, A., Khan, S., Hayat, M., Khan, F.S., Yang, M.H.: Restormer: Efficient transformer for high-resolution image restoration. In: CVPR (2022) 6

    work page 2022

  40. [40]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Zbontar, J., LeCun, Y.: Computing the stereo matching cost with a convolu- tional neural network. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 1592–1599 (2015) 1

    work page 2015

  41. [41]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Zhang, F., Prisacariu, V., Yang, R., Torr, P.H.: Ga-net: Guided aggrega- tion net for end-to-end stereo matching. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 185–194 (2019) 1

    work page 2019

  42. [42]

    In: Proceedings of the AAAI conference on artificial intelligence

    Zhang, Y., Chen, Y., Bai, X., Yu, S., Yu, K., Li, Z., Yang, K.: Adaptive unimodal cost volume filtering for deep stereo matching. In: Proceedings of the AAAI conference on artificial intelligence. vol. 34, pp. 12926–12934 (2020) 4

    work page 2020

  43. [43]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Zhao, H., Zhou, H., Zhang, Y., Chen, J., Yang, Y., Zhao, Y.: High-frequency stereo matching network. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 1327–1336 (2023) 4, 5, 10, 11

    work page 2023