Pith. sign in

REVIEW 3 major objections 7 minor 51 references

Mono2Stereo: Monocular Knowledge Transfer for Enhanced Stereo Matching

T0 review · 3 major / 7 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper argues that stereo matching networks can be made more accurate and better at generalizing to unseen scenes by transferring knowledge from monocular depth estimation, which has advanced further thanks to large-scale training…

desk verdict Solid stereo training pipeline with real zero-shot gains, but the S2DKD loss is underspecified as written and the official-benchmark gain is tiny; worth reviewing with major revision. read the letter →

arxiv 2411.09151 v1 pith:WTC6NG5M submitted 2024-11-14 cs.CV

Yuran Wang , Yingping Liang , Hesong Li , Ying Fu This is my paper · ORCID
classification cs.CV
keywords stereomatchingmonoculardepthestimationknowledgedistillationdatagenerationpre-trainingfine-tuningedge-awareinpaintingdisparity
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that stereo matching networks can be made more accurate and better at generalizing to unseen scenes by transferring knowledge from monocular depth estimation, which has advanced further thanks to large-scale training data. To show this, the authors build a two-stage training procedure: first they synthesize a large stereo training set, called DiffMFS, from ordinary monocular images by warping them according to monocular depth predictions and filling the resulting holes with an edge-aware inpainting module; then, when fine-tuning on real-world data with sparse LiDAR labels, they add a Sparse-to-Dense Knowledge Distillation (S2DKD) loss that encourages the stereo network's disparity distribution to match the dense, relative depth predicted by the monocular model. The paper reports that models pre-trained on DiffMFS and fine-tuned with S2DKD achieve lower end-point error and fewer outliers on KITTI and ETH3D than the same networks trained on SceneFlow data and standard sparse-label fine-tuning. If correct, the work would give stereo matching a practical way to benefit from the recent progress in monocular depth estimation without changing the stereo network architecture.

What carries the argument

The machinery is a two-stage training pipeline. Stage one is DiffMFS, a data generation framework: a monocular depth model (DepthAnything) predicts relative disparity, which is randomly scaled to pixel disparity; forward warping produces a right-view image, and an Edge-Aware inpainting module—where a few background edge pixels are preserved before Stable Diffusion fills the holes—produces realistic right views. Stage two is S2DKD, a loss that computes the Kullback-Leibler divergence $\mathrm{KL}(D_{\mathrm{out}}, D_{\mathrm{mono}})$ between the stereo network's predicted disparity distribution and the monocular depth distribution at randomly sampled pixels, combined with the sparse ground-truth loss to fine-tune on real data. The KL divergence is chosen specifically because it is scale-insensitive compared to L2 or gradient losses, which fail when matching metric disparity to affine-invariant relative depth.

What would settle it

An experiment that would settle the claim: fine-tune the same network with S2DKD but replace the monocular depth predictions with randomized depth maps of the same scale (or a constant map), holding all other settings fixed. If the S2DKD gain persists, the loss is not actually transferring monocular knowledge; if the gain disappears, the transfer is real. A second check is to compute the KL loss with and without the paper's implicit normalization and verify that training diverges or stalls when the normalization is removed.

Watch

Extended reading notes

Core claim

The central claim is that monocular depth models, despite predicting only scale-ambiguous relative depth, contain dense and reliable edge and structure information that stereo networks trained on sparse LiDAR labels lack, and that this information can be transferred through two mechanisms: (1) pre-training on stereo pairs synthesized from monocular images with edge-aware inpainting, and (2) fine-tuning with a KL-divergence-based Sparse-to-Dense Knowledge Distillation loss that aligns the stereo network's disparity distribution with the monocular model's relative depth distribution. The authors demonstrate this on three stereo architectures (PSMNet, CFNet, IGEV), showing consistent zero-shot improvements over SceneFlow pre-training and better fine-tuned in-domain performance on KITTI benchmarks, including an EPE of 1.05 and D1 of 4.50 on the KITTI 15 test set under a protocol that uses only the images, not the ground truth.

Load-bearing premise

The S2DKD loss assumes that a per-pixel KL divergence between the stereo network's metric disparity and the monocular model's relative depth is a meaningful training signal, even though the paper never specifies how these scalars are converted into comparable probability distributions; the L2 and gradient variants fail on exactly this scale mismatch, so the success of the KL variant depends on an unstated normalization that could be brittle across domains.

Editorial extensions

If this is right

  • Stereo networks can be pre-trained on large collections of single images instead of requiring synthetic rendered scenes, reducing reliance on computer graphics and its domain gap.
  • The dense pseudo-labels from monocular depth provide supervision in regions where LiDAR gives no signal, such as sky, distant objects, and thin structures, potentially fixing a known overfitting failure in stereo fine-tuning.
  • The approach is architecture-agnostic: PSMNet, CFNet, and IGEV all improve, so the training strategy can be applied to future stereo backbones without modifying them.
  • Because the monocular model can be swapped (DepthAnything outperforms MiDaS in the ablation), stereo matching performance should keep improving as monocular depth estimation advances.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the KL-divergence alignment is genuinely scale-insensitive, the same two-stage recipe could transfer knowledge from any relative-depth predictor (including self-supervised monocular models) to any metric task network beyond stereo, such as optical flow or multi-view depth.
  • The edge-aware inpainting module hints at a more general principle: when generative inpainting is used to complete warped images, preserving a few real background edge pixels before filling gives the diffusion model a spatial anchor, reducing foreground-background blending; this could apply to other novel-view-synthesis pipelines.
  • The paper shows gains in areas where ground truth is absent but cannot be measured on the KITTI benchmark; a natural extension is to evaluate on dense ground-truth datasets or with human judgments in those regions to quantify the claimed edge and sky improvements.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 7 minor

Summary. The paper proposes Mono2Stereo, a two-stage training framework that transfers knowledge from a pretrained monocular depth network (DepthAnything) to stereo matching networks. In the first stage, the authors generate a synthetic stereo dataset (DiffMFS) from monocular images by scaling monocular disparity maps, forward-warping to synthesize a right view, and filling occlusion holes with an edge-aware inpainting module based on Stable Diffusion. In the second stage, they introduce a Sparse-to-Dense Knowledge Distillation (S2DKD) loss that is claimed to align the stereo network's predicted disparity distribution with dense monocular depth estimates during fine-tuning on sparse LiDAR labels. Experiments are reported on ETH3D, KITTI 12, and KITTI 15 for three architectures (PSMNet, CFNet, IGEV), including zero-shot comparisons and a fine-tuned KITTI 15 test benchmark submission.

Significance. If the two-stage pipeline is reproducible as described, it offers a practical recipe for enlarging stereo training data from monocular images and for densifying supervision beyond sparse LiDAR labels, with cross-architecture validation. The zero-shot experiments in Table 1 are encouraging and suggest that the DiffMFS generation pipeline is a useful alternative to SceneFlow pretraining. The paper also ships a clear ablation structure (Table 3) for the inpainting module, the loss form, and the choice of monocular teacher. However, the significance is conditional on resolving the specification issues below: the S2DKD loss in Eq. (5) is not a well-defined KL divergence as written, and the unsupervised comparison protocol in Table 2 has a potential indirect-label leakage risk through the monocular teacher that is not discussed.

major comments (3)
  1. [Method, 'Disparity Distribution Distillation', Eqs. (4)-(5)] The KL loss in Eq. (5) is not mathematically defined as written: KL divergence is defined between probability distributions, but D_out(i,j) and D_mono(i,j) are scalar disparity values. The paper does not state how these scalars are converted to distributions (e.g., per-pixel softmax over a disparity grid, binning, or per-image min-max normalization with a temperature). Because Table 3 shows that the L2 and gradient variants fail due to scale mismatch, the KL variant must rely on an unreported alignment mechanism. The text's statement that KL is 'not sensitive to size' is not a property of KL divergence, so the actual mechanism (possibly rank-based or normalizing each image) needs to be described precisely. Without this specification, the central fine-tuning contribution is not reproducible. Please provide the exact normalization, the number of sampled pixels, and the complete per-pixel loss formula.
  2. [Experiments, 'Comparison with Unsupervised Methods', Table 2] The claim of a fair unsupervised comparison on KITTI 15 is not established. The text says the model 'is trained on our FA-Flow Dataset' - a dataset that is never defined, presumably a typo for DiffMFS. More importantly, both the pre-training disparities and the S2DKD pseudo-labels are generated by DepthAnything, and the paper does not disclose whether DepthAnything was trained on or evaluated on KITTI imagery and depth. If DepthAnything had access to KITTI ground-truth depth, the method would have indirect access to KITTI labels, invalidating the comparison with methods that truly use no KITTI labels. Please clarify DepthAnything's exposure to KITTI (and to ETH3D for the zero-shot claims), and ideally rerun the unsupervised protocol with a teacher that provably excludes these evaluation domains from its training set.
  3. [Method, overall two-stage pipeline] Both training stages align the stereo network with the same external teacher (DepthAnything): the DiffMFS pre-training uses DepthAnything's relative disparity as the supervision signal, and the fine-tuning stage uses DepthAnything's pseudo-labels in the S2DKD loss. The stereo network is therefore aligned twice to the same teacher's inverse-depth distribution, with no independent geometric or metric source in the second stage. As a result, the improvements on regions not covered by LiDAR (e.g., the sky and distant objects, as acknowledged in supplemental section A2) may reflect consistent regression toward the teacher's bias rather than improved metric disparity. This does not by itself invalidate the method, since knowledge transfer is the stated goal, but it weakens the generalization claim. Please add an experiment that uses different teachers in the two stages (e.g., MiDaS for generation and DepthAnything for distillation, and the reverse) to show that the fine-tuning gain is not simply a monotone consequence of matching the same teacher twice.
minor comments (7)
  1. [Section headings and table text] The section heading 'Spare-to-Dense Knowledge Distillation' appears in several places and should be 'Sparse-to-Dense'.
  2. [Experiments, 'Comparison with Unsupervised Methods'] The phrase 'trained on our FA-Flow Dataset' should be corrected to 'DiffMFS Dataset' to match the name used elsewhere.
  3. [Datasets section] There is a typo: 'ScaneFlow' should be 'SceneFlow'.
  4. [Eq. (3) and surrounding notation] The notation for the scaled monocular disparity is inconsistent: the text uses D'_mono, while Eq. (3) uses D_mono' in the edge mask definition; please unify the notation.
  5. [Supplemental Table 2] The PSNR column is labeled with a downward arrow, but higher PSNR is better and the reported values increase with the proposed method; the arrow should be upward.
  6. [Figure 6 caption] The caption says 'computational advantage of your approach'; it should be 'our approach'.
  7. [Conclusion and Table 3] The claim of 'significant improvements' in in-domain performance should be tempered: the official KITTI 15 D1-all improvement is 0.01 (1.59 to 1.58) in the supplemental table, and the ablation improvement of the KL loss over S2DKD-off is 0.062 in D1 on a 20% validation split; please report multiple runs or confidence intervals to support the significance claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the monocular teacher is external to this paper and all claimed improvements are measured against held-out LiDAR ground truth.

full rationale

The two-stage pipeline uses DepthAnything, an external monocular depth model, to generate pretraining disparity targets (Eq. 2: D'_mono = f * D_mono) and to provide dense pseudo-labels for the S2DKD loss (Eqs. 4-5). This is a standard knowledge-distillation setup: the stereo network is trained to imitate the teacher, but the paper's claims of zero-shot generalization and in-domain accuracy are evaluated on KITTI 2012/2015 and ETH3D ground-truth disparities, which are independent of the teacher's outputs. The fact that the same teacher appears in both stages creates a risk of bias amplification, but it does not make the evaluated result equivalent to the input by construction: the student could fail to match real LiDAR even if it perfectly imitates the teacher. The only self-citations (Liang et al. 2023; Guo et al. 2024b) appear in a related-work enumeration and are not load-bearing for any derivation. The under-specified KL normalization in Eq. 5 is a reproducibility and correctness concern, not a circularity: the loss is undefined as written, but an undefined loss cannot be a self-referential reduction. The supplemental's admission that official KITTI 15 gains are small (1.59 to 1.58 D1) is an honest limitation and further evidence that the claimed improvements are not forced by construction.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The method depends on two large external models (DepthAnything and Stable Diffusion) and on several unreported hyperparameters. The least-supported load-bearing element is the KL distribution-matching loss: the paper never explains how scalar disparity values become distributions, and the ablation shows that scale-sensitive losses fail, so the KL variant is doing unstated alignment work.

free parameters (5)
  • Random disparity scaling range [d_min, d_max]
    Eq. 2 samples f from U(d_min, d_max). This controls the disparity magnitude of every generated training pair and is central to the pretraining data.
  • Edge mask threshold tau
    Eq. 3 uses tau to detect object edges in the disparity gradient. Not reported.
  • Number of preserved boundary pixels for EA inpainting = 2 pixels
    Supplement B1 states the EA pixel movement is 2 pixels. Chosen by hand.
  • KL loss weight alpha
    Eq. 6 weights the distillation loss against sparse GT. Not reported.
  • Number of randomly sampled pixels for KL supervision
    The method randomly samples pixels for KL; the count is not specified.
assumptions (4)
  • domain assumption DepthAnything's relative inverse depth can be rescaled by a random factor into a disparity field suitable for warping.
    Used in Eq. 2 to produce D'_mono from D_mono. If the monocular depth is not geometrically consistent across the image, the synthetic right views will be distorted.
  • domain assumption Forward warping plus inpainting yields right-view images that respect stereo epipolar geometry and are realistic enough to train a stereo network.
    The entire pretraining dataset rests on this assumption; the paper validates it only indirectly via downstream stereo accuracy.
  • ad hoc to paper KL divergence between per-pixel disparity values or their local distributions is a sensible distribution-matching loss.
    Eq. 5 writes L_KL = sum of per-pixel KL terms without defining the distributions or normalization. This is a load-bearing modeling choice for S2DKD.
  • domain assumption Sparse LiDAR ground truth and dense monocular pseudo-depth can be combined in one weighted loss without conflicts.
    Eq. 6 sums L_sparse and alpha * L_KL; if the two targets contradict at a pixel, the balance is set by an unspecified alpha.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mono2Stereo: Monocular Knowledge Transfer for Enhanced Stereo Matching." pith.science (2026). https://pith.science/paper/WTC6NG5M

@misc{pith2026241109151,
  author       = {Pith},
  title        = {Pith review of: Mono2Stereo: Monocular Knowledge Transfer for Enhanced Stereo Matching},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WTC6NG5M}},
  note         = {Machine review of arXiv:2411.09151}
}
read the original abstract

The generalization and performance of stereo matching networks are limited due to the domain gap of the existing synthetic datasets and the sparseness of GT labels in the real datasets. In contrast, monocular depth estimation has achieved significant advancements, benefiting from large-scale depth datasets and self-supervised strategies. To bridge the performance gap between monocular depth estimation and stereo matching, we propose leveraging monocular knowledge transfer to enhance stereo matching, namely Mono2Stereo. We introduce knowledge transfer with a two-stage training process, comprising synthetic data pre-training and real-world data fine-tuning. In the pre-training stage, we design a data generation pipeline that synthesizes stereo training data from monocular images. This pipeline utilizes monocular depth for warping and novel view synthesis and employs our proposed Edge-Aware (EA) inpainting module to fill in missing contents in the generated images. In the fine-tuning stage, we introduce a Sparse-to-Dense Knowledge Distillation (S2DKD) strategy encouraging the distributions of predictions to align with dense monocular depths. This strategy mitigates issues with edge blurring in sparse real-world labels and enhances overall consistency. Experimental results demonstrate that our pre-trained model exhibits strong zero-shot generalization capabilities. Furthermore, domain-specific fine-tuning using our pre-trained model and S2DKD strategy significantly increments in-domain performance. The code will be made available soon.

Figures

Figures reproduced from arXiv: 2411.09151 by the authors.

Figure 1
Figure 1. Comparison between Monocular Depth and Stereo [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall architecture of our proposed method consists of two main stages: (A) Pre-training stage with a data [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Comparison between disparity from KITTI and [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: Qualitative results of IGEV trained with and without our proposed S2DKD strategy and our generated DiffMFS [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Performance on KITTI 15 with different number [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 1
Figure 1. Figure 1: Qualitative results of IGEV on KITTI testset trained with and without our proposed S2DKD strategy and our gen [PITH_FULL_IMAGE:figures/full_fig_p011_1.png]
Figure 2
Figure 2. Figure 2: Schematic diagram of the algorithm flow of different inpainting methods. This figure shows that our inpainting method [PITH_FULL_IMAGE:figures/full_fig_p012_2.png]
Figure 3
Figure 3. Figure 3: Visulization results of inpainted right-view images with or without our method. The results without our method have [PITH_FULL_IMAGE:figures/full_fig_p014_3.png]
Figure 4
Figure 4. Figure 4: Zero-shot qualitative results of IGEV on KITTI dataset trained with and without our generated DiffMFS dataset. The [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Qualitative results of IGEV on KITTI dataset finetuned with and without our proposed S2DKD strategy. The default [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 41 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Chang, J.-R.; and Chen, Y.-S. 2018. Pyramid stereo matching network. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 5410--5418

    work page 2018

  4. [4]

    Chen, W.; Fu, Z.; Yang, D.; and Deng, J. 2016. Single-image depth perception in the wild. Advances in Neural Information Processing Systems (NIPS), 29

    work page 2016

  5. [5]

    Chi, C.; Wang, Q.; Hao, T.; Guo, P.; and Yang, X. 2021. Feature-level collaboration: Joint unsupervised learning of optical flow, stereo depth and camera motion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2463--2473

    work page 2021

  6. [6]

    Cho, J.; Min, D.; Kim, Y.; and Sohn, K. 2021. Diml/cvl rgb-d dataset: 2m rgb-d images of natural indoor and outdoor scenes. arXiv preprint arXiv:2110.11590

    arXiv 2021

  7. [7]

    Eigen, D.; Puhrsch, C.; and Fergus, R. 2014. Depth map prediction from a single image using a multi-scale deep network. Advances in Neural Information Processing Systems (NeurIPS), 27

    2014

  8. [8]

    Geiger, A.; Lenz, P.; and Urtasun, R. 2012. Are we ready for autonomous driving? the kitti vision benchmark suite. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 3354--3361

    work page 2012

Show all 51 references
  1. [9]

    Godard, C.; Mac Aodha, O.; and Brostow, G. J. 2017. Unsupervised monocular depth estimation with left-right consistency. In Proceedings of the IEEE conference on computer vision and pattern recognition, 270--279

    2017

  2. [10]

    Guo, W.; Liang, Y.; Pan, Z.; Xi, Z.; Feng, J.; and Zhou, J. 2024 a . Camera-LiDAR Cross-modality Gait Recognition. arXiv preprint arXiv:2407.02038

    2024 arXiv

  3. [11]

    Guo, W.; Pan, Z.; Liang, Y.; Xi, Z.; Zhong, Z.; Feng, J.; and Zhou, J. 2024 b . LiDAR-based Person Re-identification. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 17437--17447

    2024

  4. [12]

    Guo, X.; Li, H.; Yi, S.; Ren, J.; and Wang, X. 2018. Learning monocular depth by distilling cross-domain stereo networks. In Proceedings of the European conference on computer vision (ECCV), 484--500

    2018

  5. [13]

    Guo, X.; Lu, J.; Zhang, C.; Wang, Y.; Duan, Y.; Yang, T.; Zhu, Z.; and Chen, L. 2023. Openstereo: A comprehensive benchmark for stereo matching and strong baseline. arXiv preprint arXiv:2312.00343

    2023 arXiv

  6. [14]

    Guo, X.; Yang, K.; Yang, W.; Wang, X.; and Li, H. 2019. Group-wise correlation stereo network. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 3273--3282

    2019

  7. [15]

    Kendall, A.; Martirosyan, H.; Dasgupta, S.; Henry, P.; Kennedy, R.; Bachrach, A.; and Bry, A. 2017. End-to-end learning of geometry and context for deep stereo regression. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 66--75

    2017

  8. [16]

    Lai, H.-Y.; Tsai, Y.-H.; and Chiu, W.-C. 2019. Bridging stereo matching and optical flow via spatiotemporal correspondence. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 1890--1899

    2019

  9. [17]

    Li, Z.; and Snavely, N. 2018. MegaDepth: Learning Single-View Depth Prediction from Internet Photos. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR)

    2018

  10. [18]

    Liang, Y.; Liu, J.; Zhang, D.; and Fu, Y. 2023. Mpi-flow: Learning realistic optical flow with multiplane images. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 13857--13868

    2023

  11. [19]

    Lin, T.-Y.; Maire, M.; Belongie, S.; Hays, J.; Perona, P.; Ramanan, D.; Doll \'a r, P.; and Zitnick, C. L. 2014. Microsoft coco: Common objects in context. In The European Conference on Computer Vision (ECCV), 740--755

    2014

  12. [20]

    Lipson, L.; Teed, Z.; and Deng, J. 2021. Raft-stereo: Multilevel recurrent field transforms for stereo matching. In 2021 International Conference on 3D Vision (3DV), 218--227. IEEE

    2021

  13. [21]

    R.; and Xu, J

    Liu, P.; King, I.; Lyu, M. R.; and Xu, J. 2020. Flow2stereo: Effective self-supervised learning of optical flow and stereo matching. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 6648--6657

    2020

  14. [22]

    Mayer, N.; Ilg, E.; Hausser, P.; Fischer, P.; Cremers, D.; Dosovitskiy, A.; and Brox, T. 2016. A large dataset to train convolutional networks for disparity, optical flow, and scene flow estimation. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 4040--4048

    2016

  15. [23]

    Menze, M.; and Geiger, A. 2015. Object scene flow for autonomous vehicles. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 3061--3070

    2015

  16. [24]

    Neuhold, G.; Ollmann, T.; Rota Bulo, S.; and Kontschieder, P. 2017. The mapillary vistas dataset for semantic understanding of street scenes. In the IEEE International Conference on Computer Vision, 4990--4999

    2017

  17. [25]

    Orb, M.-A. R. T. J. 2017. SLAM2: an open-source SLAM system for monocular, stereo, and RGB-D cameras. IEEE Transactions on Robotics, 33(5): 1255--1262

    2017

  18. [26]

    Ranftl, R.; Lasinger, K.; Hafner, D.; Schindler, K.; and Koltun, V. 2020. Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 44(3): 1623--1637

    2020

  19. [27]

    Ranftl, R.; Lasinger, K.; Hafner, D.; Schindler, K.; and Koltun, V. 2022. Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-Shot Cross-Dataset Transfer. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 44(3)

    2022

  20. [28]

    L.; Galliani, S.; Sattler, T.; Schindler, K.; Pollefeys, M.; and Geiger, A

    Schops, T.; Schonberger, J. L.; Galliani, S.; Sattler, T.; Schindler, K.; Pollefeys, M.; and Geiger, A. 2017. A multi-view stereo benchmark with high-resolution images and multi-camera videos. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 3260--3269

    2017

  21. [29]

    Shen, Z.; Dai, Y.; and Rao, Z. 2021. Cfnet: Cascade and fused cost volume for robust stereo matching. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 13906--13915

    2021

  22. [30]

    Tosi, F.; Tonioni, A.; De Gregorio, D.; and Poggi, M. 2023. Nerf-supervised deep stereo. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 855--866

    2023

  23. [31]

    Z.; Daniele, A

    Vasiljevic, I.; Kolkin, N.; Zhang, S.; Luo, R.; Wang, H.; Dai, F. Z.; Daniele, A. F.; Mostajabi, M.; Basart, S.; Walter, M. R.; et al. 2019. Diode: A dense indoor and outdoor depth dataset. arXiv preprint arXiv:1908.00463

    2019 arXiv

  24. [32]

    Wang, Q.; Zheng, S.; Yan, Q.; Deng, F.; Zhao, K.; and Chu, X. 2021. Irs: A large naturalistic indoor robotics stereo dataset to train deep models for disparity and surface normal estimation. In IEEE International Conference on Multimedia and Expo (ICME), 1--6

    2021

  25. [33]

    M.; Turmukhambetov, D.; Brostow, G

    Watson, J.; Aodha, O. M.; Turmukhambetov, D.; Brostow, G. J.; and Firman, M. 2020. Learning stereo from single images. In The European Conference on Computer Vision (ECCV), 722--740

    2020

  26. [34]

    Xian, K.; Zhang, J.; Wang, O.; Mai, L.; Lin, Z.; and Cao, Z. 2020. Structure-guided ranking loss for single image depth prediction. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 611--620

    2020

  27. [35]

    Xu, G.; Wang, X.; Ding, X.; and Yang, X. 2023. Iterative geometry encoding volume for stereo matching. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 21919--21928

    2023

  28. [36]

    Yang, A.; Zhang, C.; Chen, Y.; Zhuansun, Y.; and Liu, H. 2019 a . Security and privacy of smart home systems based on the Internet of Things and stereo matching algorithms. IEEE Internet of Things Journal, 7(4): 2521--2530

    2019

  29. [37]

    Yang, G.; Song, X.; Huang, C.; Deng, Z.; Shi, J.; and Zhou, B. 2019 b . Drivingstereo: A large-scale dataset for stereo matching in autonomous driving scenarios. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 899--908

    2019

  30. [38]

    Yang, G.; Zhao, H.; Shi, J.; Deng, Z.; and Jia, J. 2018. Segstereo: Exploiting semantic information for disparity estimation. In Proceedings of the European conference on computer vision (ECCV), 636--651

    2018

  31. [39]

    Yang, L.; Kang, B.; Huang, Z.; Xu, X.; Feng, J.; and Zhao, H. 2024. Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 10371--10381

    2024

  32. [40]

    Yao, Y.; Luo, Z.; Li, S.; Zhang, J.; Ren, Y.; Zhou, L.; Fang, T.; and Quan, L. 2020. Blendedmvs: A large-scale dataset for generalized multi-view stereo networks. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 1790--1799

    2020

  33. [41]

    Yuan, S.; and Tomasi, C. 2023. Ufd-prime: Unsupervised joint learning of optical flow and stereo depth through pixel-level rigid motion estimation. arXiv preprint arXiv:2310.04712

    2023 arXiv

  34. [42]

    Zhang, F.; Prisacariu, V.; Yang, R.; and Torr, P. H. 2019. Ga-net: Guided aggregation net for end-to-end stereo matching. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 185--194

    2019

  35. [43]

    H.; Lim, J.; and Suh, I

    Zhang, G.; Lee, J. H.; Lim, J.; and Suh, I. H. 2015. Building a 3-D line-based map using stereo SLAM. IEEE Transactions on Robotics, 31(6): 1364--1377

    2015

  36. [44]

    Zhou, B.; Zhao, H.; Puig, X.; Fidler, S.; Barriuso, A.; and Torralba, A. 2017. Scene parsing through ade20k dataset. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 633--641

    2017

  37. [45]

    Cheng, X.; Zhong, Y.; Harandi, M.; Dai, Y.; Chang, X.; Li, H.; Drummond, T.; and Ge, Z. 2020. Hierarchical neural architecture search for deep stereo matching. Advances in neural information processing systems, 33: 22158--22169

    2020

  38. [46]

    Li, J.; Wang, P.; Xiong, P.; Cai, T.; Yan, Z.; Yang, L.; Liu, J.; Fan, H.; and Liu, S. 2022. Practical stereo matching via cascaded recurrent network with adaptive correlation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 16263--16272

    2022

  39. [47]

    Song, X.; Zhao, X.; Fang, L.; Hu, H.; and Yu, Y. 2020. Edgestereo: An effective multi-task learning network for stereo matching and edge detection. International Journal of Computer Vision (IJCV), 128(4): 910--930

    2020

  40. [48]

    Tankovich, V.; Hane, C.; Zhang, Y.; Kowdle, A.; Fanello, S.; and Bouaziz, S. 2021. Hitnet: Hierarchical iterative tile refinement network for real-time stereo matching. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 14362--14372

    2021

  41. [49]

    Xu, G.; Cheng, J.; Guo, P.; and Yang, X. 2022. Attention concatenation volume for accurate and efficient stereo matching. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 12981--12990

    2022

  42. [50]

    Zhang, F.; Chen, Y.; Li, Z.; Hong, Z.; Liu, J.; Ma, F.; Han, J.; and Ding, E. 2019 a . Acfnet: Attentional class feature network for semantic segmentation. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 6798--6807

    2019

  43. [51]

    Zhang, F.; Prisacariu, V.; Yang, R.; and Torr, P. H. 2019 b . Ga-net: Guided aggregation net for end-to-end stereo matching. In the IEEE conference on Computer Vision and Pattern Recognition (CVPR), 185--194

    2019

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.