Superpixel Cost Volume Excitation for Stereo Matching

Junyu Dong; Lin Qi; Liyi Xu; Shanglong Liu; Wenxiang Gu

REVIEW 5 major objections 6 minor 24 references

Superpixel Cost Volume Excitation for Stereo Matching

T0 review · 5 major / 6 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 Superpixel-level supervision of the disparity probability volume improves boundary accuracy in cost-volume stereo networks at zero inference overhead.

desk verdict Nice idea, but the headline EPE drops are confounded by an unablated switch to top-k soft-argmin; the superpixel head's contribution is not yet established. read the letter →

arxiv 2411.13105 v1 pith:Q62YJX42 submitted 2024-11-20 cs.CV

Shanglong Liu , Lin Qi , Junyu Dong , Wenxiang Gu , Liyi Xu This is my paper · ORCID

classification cs.CV

keywords stereomatchingdisparityestimationsuperpixelcostvolumecross-entropylossboundaryaccuracyprobabilitytraining-onlyauxiliarybranch

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 claims that stereo matching networks built on cost volumes—4D arrays of matching scores over disparity candidates—can be made more accurate at object boundaries by supervising the disparity probability volume at the superpixel level. The authors add a superpixel segmentation branch that runs only during training, pool the network's per-pixel disparity probabilities inside each superpixel, and train the pooled distribution against an adaptive unimodal target modeled with a Laplace distribution. Because the branch is discarded at inference, the accuracy gain comes without extra computation or memory in deployment. On Scene Flow, adding this training head to GwcNet lowers end-point error from 0.76 to 0.59 pixels, with the largest visible improvements at boundaries; the same head also improves PSMNet, MobileStereo, and PCWNet. If the claim holds, it offers a cheap, transferable way to fix a known weakness of cost-volume stereo: over-smoothed and multi-peaked disparity distributions at edges.

What carries the argument

The load-bearing object is the superpixel probability volume $P_s$: for each superpixel $s$, the per-pixel disparity probabilities $C_{\mathrm{prob}}(p)$ are pooled by geometric mean in log space, $\ln(P_s) = \frac{1}{n}\sum_{p \in \tilde{m}_s} \ln(C_{\mathrm{prob}}(p))$, giving a single disparity distribution per region. This pooled distribution is compared, via cross-entropy, with a superpixel-level ground truth obtained by pooling an adaptive Laplace target $P^{\mathrm{gt}}(d) = \operatorname{softmax}(-|d - d_{\mathrm{gt}}|/v)$, where $v$ is read from the model's own aggregated volume. Two supporting mechanisms carry the claim: the superpixel guided channel excitation (SGCE) module, which multiplies the 4D cost volume by sigmoid-transformed multi-scale superpixel features along the channel dimension before each aggregation stage, and the disparity reconstruction loss $L_{\mathrm{recon}}$ that keeps the segmentation branch aligned with disparity. The entire superpixel machinery is dropped at inference, leaving the base stereo network unchanged.

What would settle it

Train the same cost-volume network with the superpixel cross-entropy loss but with a fixed, non-adaptive variance in the Laplace ground truth; if the boundary-error reductions on Scene Flow vanish, the adaptive variance computed from the model's own volume is the essential component, and if they persist, the superpixel pooling itself is what matters.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that a superpixel-level training signal can excite the latent local consistency already present in a cost volume's probability distribution. The method defines a superpixel probability volume by geometric mean pooling the per-pixel disparity probabilities inside each superpixel, and trains it to match a unimodal superpixel ground truth built from a Laplace distribution whose variance is estimated from the network's own aggregated cost volume. This superpixel cross-entropy loss, combined with a channel-excitation module that injects multi-scale superpixel features into the cost volume and a disparity-reconstruction loss that keeps superpixels disparity-aware, removes spurious modes and sharpens boundary predictions. The authors report that the auxiliary superpixel branch is needed only in training, so the resulting accuracy improvement is cost-free at inference.

Load-bearing premise

The load-bearing premise is that superpixels formed from color and spatial similarity coincide with regions of constant disparity, and that the variance read from the model's own cost volume is a reliable guide to how sharp the training target should be.

Editorial extensions

If this is right

If correct, the method gives a drop-in training-only upgrade for cost-volume stereo networks: any model that regresses disparity from a probability volume can be supervised with superpixel cross-entropy without altering deployment code.
Boundary errors, which dominate end-point error and edge metrics on Scene Flow, Middlebury, and KITTI, should shrink whenever color-consistent superpixels overlap disparity-consistent regions.
Because the loss weights in Eq. 9 are set heuristically, the approach implies a stable training regime where the same weights work across at least four different base networks.
The reported transfer to PSMNet, MobileStereo, and PCWNet indicates the benefit is not an artifact of one particular cost-volume design.
The method is explicitly not validated on iterative refinement architectures, so its scope as stated is limited to networks that regress disparity from a probabilistic cost volume.

Reading between the lines

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

Editorial extension: the same superpixel pooling mechanism could apply to any stereo pipeline that exposes a probability volume before regression, including iterative refinement models; the authors explicitly do not validate that setting, so this is an open prediction, not a result of the paper.
Editorial extension: a clean test would replace the color-derived superpixel association map with one derived from disparity or semantic labels; if boundary gains persist, the operative ingredient is regional disparity consistency rather than color affinity.
Editorial extension: the adaptive variance $v$ in Eq. 2 makes the training target depend on the network's own aggregated volume, so a controlled experiment with a fixed variance would separate the superpixel-pooling effect from the adaptive-target effect.

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.

Desk Editor's Note

read the letter

First, the new thing: a training-only head that pools the cost-volume distribution over superpixels, builds a superpixel-level ground truth from a Laplace model, and adds a cross-entropy loss plus a superpixel-guided excitation module. That combination is new relative to AcfNet, and the authors deserve credit for pushing distribution supervision to a regional level without adding inference cost. The universality experiments across PSMNet, MobileStereo, and PCWNet are the right kind of evidence.

The problem: the paper changes two variables at once. Section 3.3 replaces standard soft-argmin with top-k soft-argmin (k=6), and the baselines in Tables 2 and 3 are original GwcNet, PSMNet, etc. Nothing says those baselines also use top-k, and there is no ablation varying k. So the reported EPE drop in Table 3 (0.76 → 0.59) could come entirely from the readout change. Table 1, the ablation, also seems to use Eq. 5 in modified rows but not in the GwcNet baseline row, so it cannot separate the superpixel contributions. This is the load-bearing issue: the central claim is about Lsce and SGCE, but the evidence doesn't isolate them.

Other soft spots, in decreasing severity: the ablation table is garbled (columns Lce/Lsce/LreconC/LreconD don't line up with the text); no error bars or repeated runs; hyperparameters λ, μ, w, k are set heuristically; and the target variance v in Eq. 2 is computed from the model's own cost volume, making the supervision partly self-referential—not necessarily fatal, but it can reinforce confident errors. To the authors' credit, they do note the KITTI limitation where superpixel branches misalign with disparity-consistent regions.

Recommendation: the paper should not be accepted as-is, but it deserves a serious referee. A revision that (a) restricts the baseline to use the same top-k readout, (b) ablates k, (c) cleans up Table 1, and (d) reports variance over seeds would settle whether the superpixel head does anything real. If the confound holds and the gains vanish, it's a negative result about regional supervision; if they persist, it's a solid plug-in. Either way, a competent reviewer can sort it out.

Referee Report

5 major / 6 minor

Summary. The paper proposes a training-time superpixel module for cost-volume-based stereo matching networks. A superpixel segmentation branch (with a Superpixel Guided Channel Excitation, SGCE) modulates the cost volume, and a training head adds a superpixel cross-entropy loss (Lsce, Eq. 8) with an adaptive Laplace target (Eq. 2) plus a disparity reconstruction loss (Lrecon, Eq. 7). The claimed benefit is improved boundary accuracy and lower EPE without additional inference cost. Experiments are reported on Scene Flow, KITTI 2012/2015, and Middlebury, using GwcNet as the main baseline and integrating the training head into PSMNet, MobileStereo, and PCWNet.

Significance. The core idea—using superpixel-consistent probability distributions as a soft training constraint while keeping inference unchanged—is practical and potentially useful for cost-volume stereo networks. The training-time-only property is a genuine strength, and the authors are explicit about the limitation that the approach is designed for probabilistic cost volumes rather than iterative refinement architectures. However, the current evidence is undermined by a missing control on the disparity readout and by reporting issues. If the isolated contribution of each component can be established with a properly controlled ablation, the method would be a credible, low-cost improvement. The paper does not provide machine-checked proofs or released code; its contribution is empirical.

major comments (5)

[§3.3, Eq. (5), Table 1] The reported improvements are confounded by the unablated switch to the top-k soft-argmin readout. Eq. (5) replaces the standard full-domain soft-argmin, and §4.1 sets k=6. The manuscript never states that the GwcNet baseline or the reproduced baselines in Tables 2 and 3 use this top-k readout, and no ablation varies k. Since top-k truncates low-probability tails, it can reduce EPE independently of the proposed SGCE/Lsce/Lrecon components. Thus the ablation in Table 1 changes two variables at once and cannot attribute the gains to the superpixel mechanism. The authors should retrain the baseline with the same top-k readout and provide a k-ablation to establish the isolated contribution.
[Table 1] Table 1 is too poorly formatted to support the ablation claims. The row labeled '+ SGCE' contains no metrics, and the checkmark pattern is ambiguous: the combination '- ✓ - ✓' appears both as the second row and as the last row. It is impossible to tell which rows include SGCE and which correspond to the 'GwcNet + Ours' configuration. The table should be reformatted so each row is a uniquely named configuration with all four loss flags and the metric values, e.g., 'GwcNet', 'GwcNet + Lsce + LreconD', 'GwcNet + SGCE', etc.
[§3.2, Eq. (2)] The variance v in the Laplace target P_gt(d) is computed from the model's own aggregated cost volume (§3.2, Figure 4). This makes the sharpness of the training target depend on the model's prediction, so overconfident errors could be reinforced. This is a correctness-risk concern rather than a formal circularity, since the mode is still at the ground-truth disparity d_gt, but it deserves a direct test: compare Lsce with a detached variance (e.g., stop-gradient on v) or with a fixed v to demonstrate that the adaptive variance is necessary and does not harm.
[§4.2, Tables 1–5] No error bars, standard deviations, or repeated-run results are reported. Several ablations differ by only 0.03–0.05 EPE (e.g., 0.626 vs 0.622), and it is unclear whether these differences are significant under typical training noise. The central claim of consistent improvement would be strengthened by reporting statistics over at least three seeds.
[§4.3, Table 2] The universality study inherits the same top-k confound. The text does not state whether the reproduced baselines PSMNet, MobileStereo, and PCWNet use the top-k soft-argmin (Eq. 5) or the original full soft-argmin. Since the proposed training head includes the top-k readout, the gains in Table 2 may again reflect the readout change rather than the superpixel losses. The authors should specify the readout used in all baselines and, ideally, provide a version where the baseline is trained with the same readout.

minor comments (6)

[§4.2, Table 1] The column headers LreconC and LreconD are not defined in the text; the authors should state that one is the color-based and the other the depth-based disparity reconstruction loss, and which is which.
[§3.3, Eq. (7)] The notation for the superpixel reconstruction is unclear: define Q̃ and Q̂, and spell out how d′ and p′ are obtained; currently the sentence 'obtained by left multiplying association map Q̃ Q̂^T' is ambiguous.
[§4.3] The text refers to 'Table III' but the table is numbered Table 2; the cross-reference should be corrected.
[§4.1] The choice k=6 is justified only by 'superior performance observed in our prior work'; since k is a free parameter of the readout, an ablation over k is needed, at least in the revision.
[Eq. (3)] The geometric-mean pooling in Eq. (3) is unusual; a brief justification of why the geometric mean (rather than the arithmetic mean) is used for superpixel probability pooling would help the reader.
[§4.4] The KITTI tables (Tables 4 and 5) do not identify which GwcNet variant is used as the direct baseline ('GwcNet-gc' vs 'GwcNet-g'); the authors should align the baseline naming with the main comparison.

Circularity Check

1 steps flagged · score 5.0 of 10

The Lsce 'ground-truth' distribution is constructed from the model's own probability volume, making the superpixel cross-entropy term a self-comparison; the unablated top-k readout is an attribution confound but not a circular reduction.

self definitional [Section 3.2 (Eqs. 2-4) and Section 3.3 (Eq. 8)]
"As depicted in Figure 4, the variance v is computed based on the aggregated cost volume (i.e. probability volume). ... Lsce = − 1 Ns D−1X d=0 P gt s (d) · log Ps(d)"

P_gt in Eq. 2 is a Laplace distribution centered at d_gt whose variance v is derived from the model's own aggregated probability volume Cprob. Ps in Eqs. 3-4 is the geometric-mean superpixel pooling of that same Cprob. Therefore Lsce (Eq. 8) is a cross-entropy between two functions of the current model output: the 'ground-truth' target is not an independent label but a variance-scaled, superpixel-pooled reshaped version of the prediction itself. A confidently wrong boundary mode can produce a small v and a sharp target that reinforces the error; minimizing Lsce can amount to sharpening existing beliefs rather than correcting them.

full rationale

One genuine circular step exists: the superpixel cross-entropy target is built from the model's own probability volume via the variance v, so Eq. 8 compares the prediction with a transformed version of itself. This is a self-referential training design rather than an independent ground truth. However, the central empirical claims are not purely circular: disparity regression (Eq. 6) still uses annotated disparities, and all reported EPE/D1 numbers come from external Scene Flow, KITTI, and Middlebury evaluations, so those measurements are independent of the training-target construction. The top-k soft-argmin introduced in Eq. 5 and fixed at k=6 in Section 4.1 is a serious unablated confound for the attribution of gains to SGCE/Lsce/Lrecon, since baselines are not described as using the same readout; but that is an experimental-design and attribution problem, not an equation-level circularity. No load-bearing self-citation chain or uniqueness import is present, and the 'prior work' justification for k is not itself a derivation. Thus the overall circularity is partial and localized to the training target, giving a score of 5.

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

The central claim rests on a small set of hand-set hyperparameters and two domain assumptions about superpixel-disparity alignment and the Laplace model. No new physical entities are introduced.

free parameters (4)

lambda (weight for Lsce) = 1
Heuristically set in Section 4.1; controls the strength of the superpixel cross-entropy loss.
mu (weight for Lrecon) = 0.1
Heuristically set; controls the disparity reconstruction loss.
w (compactness weight in Lrecon) = 5e-3
Set to 5e-3 for 'appropriate disparity reconstruction loss contribution' (Section 4.1).
k (top-k in soft-argmin) = 6
Chosen based on 'the superior performance observed in our prior work' (Section 4.1); prior work not cited.

assumptions (4)

domain assumption Depth transitions smoothly within homologous regions; discontinuities only occur between distinct regions
Stated in the Introduction and Section 1 as the motivation for using superpixels.
domain assumption Superpixels derived from color images align with disparity-consistent regions
Assumed throughout, but the authors admit in Section 4.4 that on KITTI large scenes segmentation areas may deviate from disparity consistency.
ad hoc to paper The Laplace distribution with variance estimated from the cost volume models the true disparity distribution
Used to define the ground truth distribution P_gt in Eq. 2; the variance is model-dependent.
standard math Soft-argmin with top-k values effectively regresses disparity
Used in Eq. 5; a common technique from prior work [12].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Superpixel Cost Volume Excitation for Stereo Matching." pith.science (2026). https://pith.science/paper/Q62YJX42

@misc{pith2026241113105,
  author       = {Pith},
  title        = {Pith review of: Superpixel Cost Volume Excitation for Stereo Matching},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q62YJX42}},
  note         = {Machine review of arXiv:2411.13105}
}

read the original abstract

In this work, we concentrate on exciting the intrinsic local consistency of stereo matching through the incorporation of superpixel soft constraints, with the objective of mitigating inaccuracies at the boundaries of predicted disparity maps. Our approach capitalizes on the observation that neighboring pixels are predisposed to belong to the same object and exhibit closely similar intensities within the probability volume of superpixels. By incorporating this insight, our method encourages the network to generate consistent probability distributions of disparity within each superpixel, aiming to improve the overall accuracy and coherence of predicted disparity maps. Experimental evalua tions on widely-used datasets validate the efficacy of our proposed approach, demonstrating its ability to assist cost volume-based matching networks in restoring competitive performance.

Figures

Figures reproduced from arXiv: 2411.13105 by the authors.

**Figure 1.** Visualization of the real output distribution at boundaries on Scene Flow dataset. (a) is the input image, and its partial enlargement. (b) represents the disparity probability distribution of the superpixel belonging to the brown region. (c) and (d) show the output probability distributions of a given pixel from GwcNet and GwcNet+Ours. Our proposed methods rectify the incorrect distributions and avoid smoothness b… view at source ↗

**Figure 2.** The proposed stereo matching framework consists of a stereo matching pipeline and a sub-network for superpixel segmentation. The superpixel branch (cyan) takes the left image as input and assists the stereo branch (black). Furthermore, the effectiveness of a strong-constraint disparity filtering strategy is limited due to the coarse-grained nature of superpixel representation, which cannot refine to each disparity l… view at source ↗

**Figure 3.** Superpixel guided channel excitation module. The multi-scale short connections achieved through 2D convolutional kernels of varying sizes and strides combined with upsampling result in rich object context within the superpixel branch. 3 Methods 3.1 Superpixel Guided Channel Excitation As shown in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: The main components of the joint learning training head, which combines the output results from two branches, consist of a variance estimator for predicting matchbility and a superpixel pooling module, all driven by the cross-entropy loss function. 3.2 Superpixel Pool…

**Figure 5.** Figure 5: Qualitative comparisons of ablation study on Scene Flow test set [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: Qualitative results on the Middlebury test set compared to the top end-to-end deep learning approach ACVNet[3]. Middlebury Dataset. To assess model performance in real-world indoor scenes, we utilized the Middlebury dataset [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

**Figure 7.** Figure 7: Qualitative results on the KITTI 2012 (top) and KITTI 2015 (bottom) test set. White box highlighted the improvement of details. KITTI. To evaluate model performance in real-world driving scenarios, we employed the KITTI 2015 and KITTI 2012 datasets, both capturing cha…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 12 canonical work pages

[1]

IEEE Transactions on Pattern Analysis and Machine Intelligence, vol

Hirschmuller, H.: Stereo Processing by Semiglobal Matching and Mutual Informa- tion. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 30, no. 2, pp. 328–341 (2008)

work page 2008
[2]

In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

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, no. 7, pp. 12926–12934 (2020). http: //dx.doi.org/10.1609/aaai.v34i07.6991

work page doi:10.1609/aaai.v34i07.6991 2020
[3]

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

Xu G., Cheng J., Guo P., Yang X.: Attention Concatenation Volume for Accurate and Efficient Stereo Matching. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 12981–12990 (2022)

work page 2022
[4]

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

Chen L., Wang W., Mordohai P.: Learning the Distribution of Errors in Stereo Matching for Joint Disparity and Uncertainty Estimation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 17235– 17244 (2023)

work page 2023
[5]

IEEE Transactions on Pattern Analysis and Machine Intelligence vol

Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., S¨ usstrunk, S.: SLIC su- perpixels compared to state-of-the-art superpixel methods. IEEE Transactions on Pattern Analysis and Machine Intelligence vol. 34, no. 11, pp. 2274–2282 (2012). http://dx.doi.org/10.1109/tpami.2012.120

doi:10.1109/tpami.2012.120 2012
[6]

In: Computer Vision–ACCV 2018: 14th Asian Conference on Computer Vision, Perth, Australia, December 2–6, 2018, Revised Selected Papers, Part V, pp

Song, X., Zhao, X., Hu, H., Fang, L.: Edgestereo: A context integrated residual pyramid network for stereo matching. In: Computer Vision–ACCV 2018: 14th Asian Conference on Computer Vision, Perth, Australia, December 2–6, 2018, Revised Selected Papers, Part V, pp. 20–35. Springer (2019)

work page 2019
[7]

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

Xu, G., Wang, X., Ding, X., Yang, X.: Iterative geometry encoding volume for stereo matching. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 21919–21928 (2023)

work page 2023
[8]

In: Proceedings of the European Conference on Computer Vision (ECCV), pp

Jampani, V., Sun, D., Liu, M.-Y., Yang, M.-H., Kautz, J.: Superpixel sampling networks. In: Proceedings of the European Conference on Computer Vision (ECCV), pp. 352–368 (2018)

work page 2018

Show all 24 references

[9]

IEEE Transactions on Circuits and Systems for Video Technology vol

Li, L., Zhang, S., Yu, X., Zhang, L.: PMSC: PatchMatch-based superpixel cut for accurate stereo matching. IEEE Transactions on Circuits and Systems for Video Technology vol. 28, no. 3, pp. 679–692 (2016).http://dx.doi.org/10.1109/tcsvt. 2016.2628782

2016
[10]

In: European Conference on Computer Vision, pp

Shen, Z., Dai, Y., Song, X., Rao, Z., Zhou, D., Zhang, L.: PCW-Net: Pyramid combination and warping cost volume for stereo matching. In: European Conference on Computer Vision, pp. 280–297. Springer (2022)

2022
[11]

IEEE Transactions on Im- age Processing vol

Chen, J., Hou, J., Ni, Y., Chau, L.-P.: Accurate light field depth estimation with superpixel regularization over partially occluded regions. IEEE Transactions on Im- age Processing vol. 27, no. 10, pp. 4889–4900 (2018)

2018
[12]

In: Proceedings of the IEEE International Conference on Computer Vision, pp

Kendall, A., Martirosyan, H., Dasgupta, S., Henry, P., Kennedy, R., Bachrach, A., Bry, A.: End-to-end learning of geometry and context for deep stereo regression. In: Proceedings of the IEEE International Conference on Computer Vision, pp. 66–75 (2017). http://dx.doi.org/10.11...

2017 doi
[13]

In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

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). http://dx.doi.org/10.1109/cvpr.2018.00567 Superpixel Cost Volume Excitation for Stereo Matching 13

2018
[14]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pp

Guo, X., Yang, K., Yang, W., Wang, X., Li, H.: Group-wise correlation stereo net- work. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pp. 3273–3282 (2019). http://dx.doi.org/10.1109/cvpr.2019. 00339

2019 doi
[15]

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

Duggal, S., Wang, S., Ma, W.-C., Hu, R., Urtasun, R.: Deeppruner: Learning efficient stereo matching via differentiable patchmatch. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 4384–4393 (2019). http://dx.doi.org/10.1109/iccv.2019.00448

2019
[16]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pp

Tosi, F., Liao, Y., Schmitt, C., Geiger, A.: Smd-nets: Stereo mixture density net- works. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pp. 8942–8952 (2021). http://dx.doi.org/10.1109/cvpr46437. 2021.00883

2021
[17]

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

Chen, C., Chen, X., Cheng, H.: On the over-smoothing problem of CNN based disparity estimation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 8997–9005 (2019). http://dx.doi.org/10.1109/iccv. 2019.00909

2019
[18]

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

Yang, F., Sun, Q., Jin, H., Zhou, Z.: Superpixel segmentation with fully convolu- tional networks. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13964–13973 (2020). http://dx.doi.org/10.1109/ cvpr42600.2020.01398

2020
[19]

In: 18th International Conference on Pattern Recognition (ICPR’06), vol

Klaus, A., Sormann, M., Karner, K.: Segment-based stereo matching using be- lief propagation and a self-adapting dissimilarity measure. In: 18th International Conference on Pattern Recognition (ICPR’06), vol. 3, pp. 15–18. IEEE (2006). http://dx.doi.org/10.1109/icpr.2006.1033

2006
[20]

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

Wu, Z., Wu, X., Zhang, X., Wang, S., Ju, L.: Semantic stereo matching with pyramid cost volumes. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 7484–7493 (2019). http://dx.doi.org/10.1109/iccv. 2019.00758

2019
[21]

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

Lipson, L., Teed, Z., Deng, J.: Raft-stereo: Multilevel recurrent field transforms for stereo matching. In: 2021 International Conference on 3D Vision (3DV), pp. 218–227. IEEE (2021). http://dx.doi.org/10.1109/3dv53792.2021.00032

2021
[22]

W., Lee, S., Kweon, I

Bangunharcana, A., Cho, J. W., Lee, S., Kweon, I. S., Kim, K.-S., Kim, S.: Correlate-and-excite: Real-time stereo matching via guided cost volume excitation. In: 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 3542–3548. IEEE (2021). http:/...

2021
[23]

In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp

Shamsafar, F., Woerz, S., Rahim, R., Zell, A.: Mobilestereonet: Towards lightweight deep networks for stereo matching. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 2417–2426 (2022). http: //dx.doi.org/10.1109/wacv51458.2022.00075

2022
[24]

Journal of Visual Communication and Image Representation, vol

Ji, P., Li, J., Li, H., Liu, X.: Superpixel alpha-expansion and normal adjustment for stereo matching. Journal of Visual Communication and Image Representation, vol. 79, 103238 (2021)

2021

Pith tools

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

[1] [1]

IEEE Transactions on Pattern Analysis and Machine Intelligence, vol

Hirschmuller, H.: Stereo Processing by Semiglobal Matching and Mutual Informa- tion. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 30, no. 2, pp. 328–341 (2008)

work page 2008

[2] [2]

In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

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, no. 7, pp. 12926–12934 (2020). http: //dx.doi.org/10.1609/aaai.v34i07.6991

work page doi:10.1609/aaai.v34i07.6991 2020

[3] [3]

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

Xu G., Cheng J., Guo P., Yang X.: Attention Concatenation Volume for Accurate and Efficient Stereo Matching. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 12981–12990 (2022)

work page 2022

[4] [4]

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

Chen L., Wang W., Mordohai P.: Learning the Distribution of Errors in Stereo Matching for Joint Disparity and Uncertainty Estimation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 17235– 17244 (2023)

work page 2023

[5] [5]

IEEE Transactions on Pattern Analysis and Machine Intelligence vol

Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., S¨ usstrunk, S.: SLIC su- perpixels compared to state-of-the-art superpixel methods. IEEE Transactions on Pattern Analysis and Machine Intelligence vol. 34, no. 11, pp. 2274–2282 (2012). http://dx.doi.org/10.1109/tpami.2012.120

doi:10.1109/tpami.2012.120 2012

[6] [6]

In: Computer Vision–ACCV 2018: 14th Asian Conference on Computer Vision, Perth, Australia, December 2–6, 2018, Revised Selected Papers, Part V, pp

Song, X., Zhao, X., Hu, H., Fang, L.: Edgestereo: A context integrated residual pyramid network for stereo matching. In: Computer Vision–ACCV 2018: 14th Asian Conference on Computer Vision, Perth, Australia, December 2–6, 2018, Revised Selected Papers, Part V, pp. 20–35. Springer (2019)

work page 2019

[7] [7]

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

Xu, G., Wang, X., Ding, X., Yang, X.: Iterative geometry encoding volume for stereo matching. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 21919–21928 (2023)

work page 2023

[8] [8]

In: Proceedings of the European Conference on Computer Vision (ECCV), pp

Jampani, V., Sun, D., Liu, M.-Y., Yang, M.-H., Kautz, J.: Superpixel sampling networks. In: Proceedings of the European Conference on Computer Vision (ECCV), pp. 352–368 (2018)

work page 2018