Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

FlowSeek shows that a compact optical flow network, boosted by frozen depth-model features and classical motion bases, matches or beats state-of-the-art flow methods while training on a single consumer GPU.

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 ↗

FlowSeek integrates a frozen depth foundation model and classical motion bases into a SEA-RAFT-style flow network, achieving state-of-the-art zero-shot generalization when trained on a single GPU.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection Solid empirical architecture paper with a real single-GPU training contribution, but the depth-prior attribution is unproven and the baseline numbers need cleaning up. the 3 major comments →

arxiv 2509.05297 v1 pith:57AQOK4S submitted 2025-09-05 cs.CV

FlowSeek: Optical Flow Made Easier with Depth Foundation Models and Motion Bases

Matteo Poggi , Fabio Tosi This is my paper
classification cs.CV
keywords optical flowdepth foundation modelsmotion basescross-dataset generalizationsingle-GPU trainingSEA-RAFTzero-shot transferrigid motion subspace
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 reading

FlowSeek argues that state-of-the-art optical flow does not require a large cluster of GPUs. By injecting priors from a frozen depth foundation model (Depth Anything v2) and classical low-dimensional motion bases into a SEA-RAFT-style architecture, the authors train a compact model on a single consumer GPU and report a relative improvement of 10% on Sintel Final and 15% on KITTI over the previous state of the art, with strong zero-shot results on Spring and LayeredFlow. The paper claims the depth model's inverse-depth maps and decoder features, combined with an eight-dimensional rigid-motion basis, give the flow network a strong geometric starting point that generalizes across datasets. If right, this lowers the hardware barrier for flow research and suggests that recycling large pretrained geometric models can substitute for brute-force training scale.

Core claim

The central claim is that an optical flow network trained on one consumer GPU can outperform models trained on eight GPUs when it is seeded with geometric priors from a depth foundation model. Inverse-depth maps D0, D1 and last-decoder features from Depth Anything v2 are fused into the backbone's features and context; a set of eight motion bases — three translational and five rotational components derived from D0 with focal-length terms scaled away — is processed by a small BaseNet and injected as an initial motion prior. With the SEA-RAFT training recipe and mixture-of-Laplace supervision, FlowSeek variants beat their SEA-RAFT counterparts on TartanAir, Sintel, KITTI, Spring, and LayeredFlo

What carries the argument

Two priors carry the argument. First, a frozen depth foundation model (Depth Anything v2) supplies inverse-depth maps and decoder features that are correlated with depth; the features are adapted by a three-convolution BottleNeck and concatenated into the correlation-volume features, and the depth maps can be added to the context network. Second, from the classical result that a rigid scene's optical flow lies in a six-dimensional subspace, the paper builds eight motion bases from inverse depth, removes focal-length dependence by rescaling, and runs them through a BaseNet so the network starts from a geometrically plausible motion subspace and refines it iteratively.

Load-bearing premise

The improvement is attributed to depth priors, but no experiment swaps the depth model for a different strong feature extractor, and the motion-basis prior is assumed to help on non-rigid scenes even though it is derived for rigid motion.

What would settle it

Train FlowSeek with the depth foundation model replaced by a frozen generic dense feature extractor (e.g., a self-supervised ViT) and identical BottleNeck and BaseNet modules; if Sintel Final and KITTI errors stay within the reported margins, depth-specific priors are not the cause. Also evaluate on a clearly non-rigid sequence and compare FlowSeek with and without the BaseNet to see whether the rigid-motion initial guess hurts.

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

If this is right

  • State-of-the-art cross-dataset generalization on Sintel Final, KITTI, Spring, and LayeredFlow becomes reachable with a single consumer GPU, lowering the hardware barrier for flow research.
  • Depth foundation models can be recycled as frozen priors for motion estimation rather than trained from scratch, avoiding prohibitive compute.
  • The combination of depth features and motion bases outperforms either prior alone, and the bases beat feeding raw depth to the same network.
  • The recipe transfers across backbones: CRAFT and FlowFormer also improve when given the same priors.
  • Newer or stronger depth foundation models should translate directly into further flow accuracy gains.

Where Pith is reading between the lines

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

  • The paper does not ablate whether the gains come from depth-specific information or from any dense feature extractor; swapping in a generic self-supervised visual feature model would test whether depth is the causal ingredient.
  • The rigid-motion basis is theoretically justified only for static scenes; testing on heavily non-rigid motion would reveal whether the 'initial guess' helps or hurts when the prior is wrong.
  • The success suggests a broader recipe: freeze a large geometric foundation model and inject classical low-dimensional structure to train compact models for other dense prediction tasks on modest hardware.
  • The 8x smaller batch size changes optimization dynamics as well as memory; part of the generalization gain may come from the regularization effect of small batches, not only from the priors.
Share X Bluesky LinkedIn Reddit HN

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 / 4 minor

Summary. The paper proposes FlowSeek, an optical flow architecture built on SEA-RAFT. It enriches the correlation features with the last decoder features of a monocular depth foundation model (Depth Anything v2), optionally feeds the estimated depth maps to the context network, and injects a classical rigid-motion basis (Heeger & Jepson) through a small BasesNet. Training is performed on a single RTX 3090 with batch sizes 6/4, i.e. at roughly an 8x smaller hardware budget than the multi-GPU SEA-RAFT baselines. The paper reports ablations on TartanAir/KITTI 2012 and zero-shot generalization results on Sintel, KITTI 2015, Spring and LayeredFlow, claiming superior accuracy over SEA-RAFT variants and, in the abstract, a relative improvement of 10% and 15% over SEA-RAFT on Sintel Final and KITTI.

Significance. If the empirical results hold, the paper makes a useful practical contribution: it shows that competitive optical flow models can be trained on a single consumer GPU by reusing pre-trained geometric foundation models instead of training from scratch on a large cluster. The evaluation is comparatively broad: it includes several depth foundation models, two flow backbones (SEA-RAFT, CRAFT, FlowFormer), multiple model sizes, and four zero-shot generalization benchmarks. The one-GPU retrained SEA-RAFT baselines are an appropriate comparison point, and the paper honestly states its reliance on large pre-trained models. The main weakness is that the mechanism attributed to the improvement—depth-specific and motion-basis-specific priors—is not isolated from a generic improvement in feature quality or capacity; the headline numbers in the abstract also need to be reconciled with the tables.

major comments (3)
  1. [§3.2, Eq. (6)–(7); Table 1] The central attribution claim is not isolated. Φ in Eq. (6) is the last decoder feature of Depth Anything v2, a model built on DINOv2 features; after the BottleNeck it is concatenated to the correlation features. Table 1 varies the depth foundation model (DPT, Depth Anything v1, v2) but never replaces Φ with an equally strong non-depth feature extractor (e.g., DINOv2 or a generic dense ViT feature) at the same resolution. The observed gains from +Φ and the selected best configuration could therefore be due to stronger visual features rather than depth-specific geometric knowledge. Please add such a control, or explicitly rephrase the paper's attribution so that it claims improved features from a foundation model rather than depth priors specifically.
  2. [Abstract; Table 3] The abstract's 'relative improvement of 10 and 15%' does not match any single comparison in Table 3. With Tartan pretraining, FlowSeek (L) vs SEA-RAFT (L) as reported by the authors gives about 35% on Sintel Final (3.37→2.18) and about 12% on KITTI Fl-all (12.7→11.2); against the one-GPU retrained SEA-RAFT (L) the values are about 20% and 17%. The claimed 10%/15% figures need to be corrected or explicitly tied to a stated baseline, training schedule, and metric. It should also be clarified whether the comparison is within the Tartan-pretrained block (apples-to-apples) or against C→T-only baselines.
  3. [Table 2; §3.3] The value of the motion-basis prior is not established against a non-motion control. Table 2 only compares B_motion with the depth map D0 as the BaseNet input, both of which are geometry-based; it does not test whether the improvement comes from the rigid-motion subspace structure rather than from any arbitrary extra 8-channel input. Moreover, on KITTI 2015 dynamic regions B_motion is worse than D0 (Fl-EPE 2.60 vs 2.49; Fl-All 9.07 vs 8.61), so the benefit is limited to static regions. A control with a generic prior of matched channel count (e.g., coordinate grids or learned embeddings) would clarify the role of the motion bases.
minor comments (4)
  1. [Eq. (14)] The notation Δ_i^F is used both for a flow residual and for the density of a mixture of Laplace distributions. The equation appears to define a probability density over x, not the residual itself; please clarify the distinction between the predicted distribution parameters and the actual residual used in Eq. (4).
  2. [§4.3] Typo: 'dowsampling' should be 'downsampling' in the Spring paragraph.
  3. [Table 3] The one-GPU SEA-RAFT numbers appear to be from a single retraining run; given the unusual result that the one-GPU SEA-RAFT (S) is better on Sintel Final than the authors' multi-GPU model (3.02 vs 4.32), it would be helpful to report the variance or number of seeds used for these baselines.
  4. [Table 1; §4.2] The text describes the model as 'compact, yet accurate', but the MAC counts in Table 1 show that adding the BaseNet and Φ roughly doubles the cost of the small model (e.g., 694.7G vs 284.7G for SEA-RAFT (S)). This is not a problem for the central claim, but the wording should be qualified.

Circularity Check

0 steps flagged

No significant circularity; FlowSeek is an empirical architecture paper whose gains are benchmarked against external baselines and whose priors come from external foundation models and classical motion bases.

full rationale

FlowSeek's central claim is an empirical improvement on optical flow benchmarks (Sintel, KITTI, Spring, LayeredFlow) relative to SEA-RAFT and other published methods. The derivation chain does not define any quantity in terms of the target prediction, and no fitted parameter is renamed as a prediction. The depth foundation model (Depth Anything v2) is an external, pre-trained model; its features Φ and depth maps D are injected into the flow network via concatenation (Eqs. 6-8) and are not fitted to the flow benchmark. The motion-basis prior is a classical six/eight-dimensional parametrization (Heeger & Jepson, ref. [21]) derived from depth and camera geometry, used only to initialize/enrich context features, not to predict flow directly. The supervision is standard negative log-likelihood on flow ground truth (Eqs. 14-15), so the output is genuinely trained from data rather than being a restatement of the inputs. The self-citations ([1], [2] by the authors) are to prior stereo/flow works and are not load-bearing for the main result; they are not invoked as a uniqueness theorem or as justification for an unverified prior. The paper does not claim to mathematically derive optical flow from depth; it claims that combining these priors improves generalization, which is supported by ablations and cross-dataset tables. The skeptical concern that gains might come from generic strong features rather than depth-specific priors (no DINOv2 control) is an attribution/correctness issue, not circularity: the paper does not define its conclusion into its premises. No equation in the paper reduces to a quantity fitted by the authors, and no self-citation chain forces the central claim. Hence score 0.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 0 invented entities

The paper introduces no new physical entities. Its epistemic load is carried by the frozen depth foundation model, the classical motion-basis subspace, and the chosen training recipe. The main unstated load is that 'depth features' are the right mechanism for the gains, as opposed to any strong feature extractor.

free parameters (2)
  • Prior combination selection (Phi + BaseNet) = Depth features Phi plus motion-basis BaseNet; D not fed to ContextNet
    Table 1 shows this combination gives the best TartanAir/KITTI 2012 results; the choice is made from validation data and then fixed as the final model.
  • Iterations iters = 4 for T and M, 12 for S and L
    Sets the recurrent refinement budget; more iterations consistently improve accuracy at higher MACs. This is a standard hyperparameter, not a fitted constant.
axioms (4)
  • standard math For a static scene with known inverse depth, optical flow lies in a six-dimensional linear subspace spanned by translational and rotational bases (Heeger & Jepson).
    Invoked in Section 3.3, Eqs. 9-12, to construct B_motion; the paper extends it to eight bases by splitting rotations and treats it as an initial guess for non-rigid scenes.
  • domain assumption Depth Anything v2 provides a useful inverse-depth prior for arbitrary input images, including non-metric relative depth.
    Used throughout Section 3.2 and Table 1; the prior's scale is arbitrary, and the correlation volume and bases must tolerate relative-depth scale shifts across domains.
  • domain assumption The SEA-RAFT architecture, mixture-of-Laplace loss, and multi-stage training schedule are taken as-is from the cited SEA-RAFT codebase.
    FlowSeek's contribution is additive to this backbone; if that training recipe is flawed or mis-evaluated, the comparison in Tables 3-5 is affected.
  • ad hoc to paper Camera intrinsics can be eliminated by assuming fx=fy and principal point at image center; bases are arbitrarily scaled.
    Section 3.3, Eq. 11: this removes the need for known focal length, but the induced basis is only an approximation for real cameras with other intrinsics.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of FlowSeek: Optical Flow Made Easier with Depth Foundation Models and Motion Bases." pith.science (2026). https://pith.science/paper/57AQOK4S

@misc{pith2026250905297,
  author       = {Pith},
  title        = {Pith review of: FlowSeek: Optical Flow Made Easier with Depth Foundation Models and Motion Bases},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/57AQOK4S}},
  note         = {Machine review of arXiv:2509.05297}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We present FlowSeek, a novel framework for optical flow requiring minimal hardware resources for training. FlowSeek marries the latest advances on the design space of optical flow networks with cutting-edge single-image depth foundation models and classical low-dimensional motion parametrization, implementing a compact, yet accurate architecture. FlowSeek is trained on a single consumer-grade GPU, a hardware budget about 8x lower compared to most recent methods, and still achieves superior cross-dataset generalization on Sintel Final and KITTI, with a relative improvement of 10 and 15% over the previous state-of-the-art SEA-RAFT, as well as on Spring and LayeredFlow datasets.

Figures

Figures reproduced from arXiv: 2509.05297 by Fabio Tosi, Matteo Poggi.

Figure 1
Figure 1. Figure 1: FlowSeek in Action. State-of-the-art optical flow models struggle at generalizing across different domains, with a lack of fine details in their predictions. FlowSeek achieves superior generalization by exploiting the strong priors from depth foundation models. Abstract We present FlowSeek, a novel framework for optical flow re￾quiring minimal hardware resources for training. FlowSeek marries the latest ad… view at source ↗
Figure 2
Figure 2. Figure 2: Architecture Overview. Our proposed FlowSeek architecture processes a pair of images I0, I1 through parallel paths: a shared￾weight feature extractor produces F0, F1, while a depth foundation model (e.g. Depth Anything v2 [93]) estimates depth maps D0, D1 and features Φ0, Φ1. These are combined to obtain enriched features F Φ 0 , F Φ 1 for building 4D correlation volumes {Vs } S s . A Base Network extracts… view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative Results on Sintel [8]. From left to right: first frame, flow by SEA-RAFT (L) and FlowSeek (L), ground-truth flow. EPE: 1.232 EPE: 1.162 EPE: 1.835 EPE: 1.460 I0 SEA-RAFT (L) [81] FlowSeek (L) [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative Results on KITTI 2015 [54]. From left to right: first frame, flow by SEA-RAFT (L) and FlowSeek (L). Optical Flow Backbones. We further prove the gen￾erality of our approach by empowering two different flow backbones, respectively CRAFT [68] and FlowFormer [26], with the priors extracted by a BaseNet. This addition largely improves the accuracy of both baseline model. Input to the BaseNet. Final… view at source ↗
Figure 5
Figure 5. Figure 5: shows qualitative comparisons between the pre￾dictions by SEA-RAFT (L) and FlowSeek (L), with the lat￾ter recovering finer details and exhibiting fewer artifacts. Qualitative Results. We refer the reader to the sup￾plementary material for additional qualitative samples con￾cerning the experiments discussed so far. Method All EPE↓ 1px↓ 3px↓ 5px↓ FlowNet-C [15] 9.71 89.07 61.51 43.93 FlowNet2 [30] 10.07 77.5… view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative results on TartanAir. From left to right: first frame, flow predicted by SEA-RAFT (S) and ablated versions of FlowSeek (T). In this document, we report additional qualitative results concerning the experiments reported in the main paper. Specifically, Figures 6 and 7 provide a visual comparison between SEA-RAFT (S) and the different versions of FlowSeek (T) ablated in [PITH_FULL_IMAGE:figures/… view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative results on KITTI 2012. From left to right, on two rows: first frame, flow predicted by SEA-RAFT (S) and ablated versions of FlowSeek (T). EPE: 0.301 EPE: 0.291 EPE: 0.214 EPE: 0.168 EPE: 0.210 EPE: 0.193 I0 SEA-RAFT (S) FlowSeek (T) [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative results on Sintel. From left to right: first frame, flow by SEA-RAFT (S) and FlowSeek (T). 15 [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative results on KITTI 2015. From left to right: first frame, flow by SEA-RAFT (S) and FlowSeek (T). EPE: 0.058 EPE: 0.046 EPE: 0.084 EPE: 0.074 I0 SEA-RAFT (S) FlowSeek (T) [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Qualitative results on Spring – “C→T→TSKH” schedule after TartanAir pretraining. From left to right: first frame, flow by SEA-RAFT (S) and FlowSeek (T). Extra Data Method Spring (train) 1px↓ EPE↓ Tartan SEA-RAFT (S) 4.574 0.597 SEA-RAFT (M) 4.815 0.580 SEA-RAFT (L) 5.988 0.850 FlowSeek (T) 4.070 0.416 FlowSeek (S) 3.999 0.410 FlowSeek (M) 4.094 0.424 FlowSeek (L) 4.067 0.422 Extra Data Method Spring (trai… view at source ↗
Figure 11
Figure 11. Figure 11: Qualitative results on Spring – “C→T” schedule after TartanAir pretraining. From left to right: first frame, flow by SEA-RAFT (S) and FlowSeek (T). Method All Transparent Reflective Diffuse EPE↓ 1px↓ 3px↓ 5px↓ EPE↓ 1px↓ 3px↓ 5px↓ EPE↓ 1px↓ 3px↓ 5px↓ EPE↓ 1px↓ 3px↓ 5px↓ FlowNet-C 9.71 89.07 61.51 43.93 11.08 89.23 62.43 45.05 6.38 88.25 58.36 40.03 8.53 89.08 54.13 35.35 FlowNet2 10.07 77.56 54.22 42.13 11… view at source ↗

discussion (0)

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

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. UniRED: Unified RGB-D Video Frame Interpolation with Event Guidance

    cs.CV 2026-06 unverdicted novelty 6.0

    UniRED proposes a multimodal RGB-D video interpolation method guided by event data that extracts fused cues, refines bidirectional flow for RGB and depth, and synthesizes frames via warping, plus releases a new RGB-D-...

Reference graph

Works this paper leans on

99 extracted references · 69 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    Learning optical flow from still images

    Filippo Aleotti, Matteo Poggi, and Stefano Mattoccia. Learning optical flow from still images. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 15201–15211, 2021. 2

    2021

  2. [2]

    Stereo anywhere: Robust zero-shot deep stereo matching even where either stereo or mono fail

    Luca Bartolomei, Fabio Tosi, Matteo Poggi, and Stefano Mattoccia. Stereo anywhere: Robust zero-shot deep stereo matching even where either stereo or mono fail. InProceed- ings of the Computer Vision and Pattern Recognition Con- ference (CVPR), pages 1013–1027, 2025. 2, 3

    2025

  3. [3]

    A framework for the robust estimation of optical flow

    Michael J Black and Padmanabhan Anandan. A framework for the robust estimation of optical flow. In1993 (4th) In- ternational Conference on Computer Vision, pages 231–236. IEEE, 1993. 2

    1993

  4. [4]

    Depth pro: Sharp monocular metric depth in less than a second.arXiv preprint arXiv:2410.02073, 2024

    Aleksei Bochkovskii, Ama ¨el Delaunoy, Hugo Germain, Marcel Santos, Yichao Zhou, Stephan R Richter, and Vladlen Koltun. Depth pro: Sharp monocular metric depth in less than a second.arXiv preprint arXiv:2410.02073, 2024. 3

    Pith/arXiv arXiv 2024

  5. [5]

    Dimensions of motion: Monocular prediction through flow subspaces

    Richard Strong Bowen, Richard Tucker, Ramin Zabih, and Noah Snavely. Dimensions of motion: Monocular prediction through flow subspaces. In2022 International Conference on 3D Vision (3DV), pages 454–464, 2022. 2, 4

    2022

  6. [6]

    High accuracy optical flow estimation based on a theory for warping

    Thomas Brox, Andr ´es Bruhn, Nils Papenberg, and Joachim Weickert. High accuracy optical flow estimation based on a theory for warping. InEuropean conference on computer vision, pages 25–36. Springer, 2004. 1

    2004

  7. [7]

    Large displacement optical flow

    Thomas Brox, Christoph Bregler, and Jitendra Malik. Large displacement optical flow. In2009 IEEE Conference on Computer Vision and Pattern Recognition, pages 41–48. IEEE, 2009. 1

    2009

  8. [8]

    A naturalistic open source movie for op- tical flow evaluation

    Daniel J Butler, Jonas Wulff, Garrett B Stanley, and Michael J Black. A naturalistic open source movie for op- tical flow evaluation. InEuropean conference on computer vision, pages 611–625. Springer, 2012. 6, 7

    2012

  9. [9]

    Emerg- ing properties in self-supervised vision transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Herv ´e J´egou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerg- ing properties in self-supervised vision transformers. InPro- ceedings of the IEEE/CVF international conference on com- puter vision, pages 9650–9660, 2021. 3

    2021

  10. [10]

    Full flow: Optical flow estimation by global optimization over regular grids

    Qifeng Chen and Vladlen Koltun. Full flow: Optical flow estimation by global optimization over regular grids. InPro- ceedings of the IEEE conference on computer vision and pat- tern recognition, pages 4706–4714, 2016. 1

    2016

  11. [11]

    Monster: Marry monodepth to stereo unleashes power

    Junda Cheng, Longliang Liu, Gangwei Xu, Xianqi Wang, Zhaoxing Zhang, Yong Deng, Jinliang Zang, Yurui Chen, Zhipeng Cai, and Xin Yang. Monster: Marry monodepth to stereo unleashes power. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,

  12. [12]

    Flowtrack: Revisiting optical flow for long- range dense tracking

    Seokju Cho, Jiahui Huang, Seungryong Kim, and Joon- Young Lee. Flowtrack: Revisiting optical flow for long- range dense tracking. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 19268–19277, 2024. 2

    2024

  13. [13]

    Explicit motion disen- tangling for efficient optical flow estimation

    Changxing Deng, Ao Luo, Haibin Huang, Shaodan Ma, Jiangyu Liu, and Shuaicheng Liu. Explicit motion disen- tangling for efficient optical flow estimation. InProceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 9521–9530, 2023. 2, 6

    2023

  14. [14]

    Rethinking opti- cal flow from geometric matching consistent perspective

    Qiaole Dong, Chenjie Cao, and Yanwei Fu. Rethinking opti- cal flow from geometric matching consistent perspective. In Proceedings of the IEEE/CVF Conference on computer vi- sion and pattern recognition, pages 1337–1347, 2023. 2, 8

    2023

  15. [15]

    Flownet: Learning optical flow with convolutional networks

    Alexey Dosovitskiy, Philipp Fischer, Eddy Ilg, Philip Hausser, Caner Hazirbas, Vladimir Golkov, Patrick Van Der Smagt, Daniel Cremers, and Thomas Brox. Flownet: Learning optical flow with convolutional networks. InPro- ceedings of the IEEE international conference on computer vision, pages 2758–2766, 2015. 1, 2, 6, 8

    2015

  16. [16]

    Fast dynamic radiance fields with time-aware neural voxels

    Jiemin Fang, Taoran Yi, Xinggang Wang, Lingxi Xie, Xi- aopeng Zhang, Wenyu Liu, Matthias Nießner, and Qi Tian. Fast dynamic radiance fields with time-aware neural voxels. InSIGGRAPH Asia 2022 Conference Papers, 2022. 1

    2022

  17. [17]

    Vision meets robotics: The kitti dataset.The Inter- national Journal of Robotics Research, 32(11):1231–1237,

    Andreas Geiger, Philip Lenz, Christoph Stiller, and Raquel Urtasun. Vision meets robotics: The kitti dataset.The Inter- national Journal of Robotics Research, 32(11):1231–1237,

  18. [18]

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning.arXiv preprint arXiv:2501.12948, 2025

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning.arXiv preprint arXiv:2501.12948, 2025. 2

    Pith/arXiv arXiv 2025

  19. [19]

    Realflow: Em- based realistic optical flow dataset generation from videos

    Yunhui Han, Kunming Luo, Ao Luo, Jiangyu Liu, Haoqiang Fan, Guiming Luo, and Shuaicheng Liu. Realflow: Em- based realistic optical flow dataset generation from videos. InEuropean conference on computer vision, pages 288–305. Springer, 2022. 2

    2022

  20. [20]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. InProceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 3, 5

    2016

  21. [21]

    Subspace methods for recovering rigid motion i: Algorithm and implementation

    David J Heeger and Allan D Jepson. Subspace methods for recovering rigid motion i: Algorithm and implementation. International Journal of Computer Vision, 7:95–117, 1992. 2, 4

    1992

  22. [22]

    Determining op- tical flow.Artificial intelligence, 17(1-3):185–203, 1981

    Berthold KP Horn and Brian G Schunck. Determining op- tical flow.Artificial intelligence, 17(1-3):185–203, 1981. 1, 2

    work page 1981

  23. [23]

    Efficient coarse-to- fine patchmatch for large displacement optical flow

    Yinlin Hu, Rui Song, and Yunsong Li. Efficient coarse-to- fine patchmatch for large displacement optical flow. InPro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 5704–5712, 2016. 1

    work page 2016

  24. [24]

    Robust interpola- tion of correspondences for large displacement optical flow

    Yinlin Hu, Yunsong Li, and Rui Song. Robust interpola- tion of correspondences for large displacement optical flow. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. 1

    work page 2017

  25. [25]

    Rife: Real-time intermediate flow estimation for video frame interpolation.arXiv preprint arXiv:2011.06294, 2020

    Zhewei Huang, Tianyuan Zhang, Wen Heng, Boxin Shi, and Shuchang Zhou. Rife: Real-time intermediate flow estimation for video frame interpolation.arXiv preprint arXiv:2011.06294, 2020. 1 9

    Pith/arXiv arXiv 2011

  26. [26]

    Flowformer: A transformer architecture for optical flow

    Zhaoyang Huang, Xiaoyu Shi, Chao Zhang, Qiang Wang, Ka Chun Cheung, Hongwei Qin, Jifeng Dai, and Hongsheng Li. Flowformer: A transformer architecture for optical flow. InEuropean conference on computer vision, pages 668–685. Springer, 2022. 1, 2, 6, 7, 8

    work page 2022

  27. [27]

    LiteFlowNet3: Resolv- ing Correspondence Ambiguity for More Accurate Optical Flow Estimation

    Tak-Wai Hui and Chen Change Loy. LiteFlowNet3: Resolv- ing Correspondence Ambiguity for More Accurate Optical Flow Estimation. InProceedings of the European Confer- ence on Computer Vision (ECCV), 2020. 1

    work page 2020

  28. [28]

    Lite- FlowNet: A Lightweight Convolutional Neural Network for Optical Flow Estimation

    Tak-Wai Hui, Xiaoou Tang, and Chen Change Loy. Lite- FlowNet: A Lightweight Convolutional Neural Network for Optical Flow Estimation. InProceedings of IEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages 8981–8989, 2018. 2

    work page 2018

  29. [29]

    A lightweight optical flow cnn - revisiting data fidelity and reg- ularization

    Tak-Wai Hui, Xiaoou Tang, and Chen Change Loy. A lightweight optical flow cnn - revisiting data fidelity and reg- ularization. InIEEE Transactions on Pattern Analysis and Machine Intelligence, 2020. 1, 2

    work page 2020

  30. [30]

    E. Ilg, N. Mayer, T. Saikia, M. Keuper, A. Dosovitskiy, and T. Brox. Flownet 2.0: Evolution of optical flow estimation with deep networks. InIEEE Conference on Computer Vi- sion and Pattern Recognition (CVPR), 2017. 1, 2, 8

    work page 2017

  31. [31]

    Brostow, and Jamie Watson

    Sergio Izquierdo, Mohamed Sayed, Michael Firman, Guillermo Garcia-Hernando, Daniyar Turmukhambetov, Javier Civera, Oisin Mac Aodha, Gabriel J. Brostow, and Jamie Watson. MVSAnywhere: Zero shot multi-view stereo. InCVPR, 2025. 2

    work page 2025

  32. [32]

    Ccmr: High resolution optical flow estimation via coarse-to-fine context-guided motion reasoning

    Azin Jahedi, Maximilian Luz, Marc Rivinius, and Andr ´es Bruhn. Ccmr: High resolution optical flow estimation via coarse-to-fine context-guided motion reasoning. InProceed- ings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 6899–6908, 2024. 2

    work page 2024

  33. [33]

    Ms-raft+: high resolution multi-scale raft.International Journal of Computer Vision, 132(5): 1835–1856, 2024

    Azin Jahedi, Maximilian Luz, Marc Rivinius, Lukas Mehl, and Andr ´es Bruhn. Ms-raft+: high resolution multi-scale raft.International Journal of Computer Vision, 132(5): 1835–1856, 2024. 2, 8

    work page 2024

  34. [34]

    Distractflow: Improving optical flow estimation via real- istic distractions and pseudo-labeling

    Jisoo Jeong, Hong Cai, Risheek Garrepalli, and Fatih Porikli. Distractflow: Improving optical flow estimation via real- istic distractions and pseudo-labeling. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13691–13700, 2023. 2

    work page 2023

  35. [35]

    Defom-stereo: Depth foundation model based stereo matching

    Hualie Jiang, Zhiqiang Lou, Laiyan Ding, Rui Xu, Minglang Tan, Wenjie Jiang, and Rui Huang. Defom-stereo: Depth foundation model based stereo matching. InIEEE Interna- tional Conference on Computer Vision and Pattern Recogni- tion (CVPR), 2025. 2, 3, 4

    work page 2025

  36. [36]

    Learning to estimate hidden motions with global motion aggregation

    Shihao Jiang, Dylan Campbell, Yao Lu, Hongdong Li, and Richard Hartley. Learning to estimate hidden motions with global motion aggregation. InProceedings of the IEEE/CVF international conference on computer vision, pages 9772– 9781, 2021. 2, 6, 8

    work page 2021

  37. [37]

    Effiscene: Efficient per-pixel rigidity inference for unsupervised joint learning of optical flow, depth, camera pose and motion seg- mentation

    Yang Jiao, Trac D Tran, and Guangming Shi. Effiscene: Efficient per-pixel rigidity inference for unsupervised joint learning of optical flow, depth, camera pose and motion seg- mentation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5538– 5547, 2021. 2

    work page 2021

  38. [38]

    What mat- ters in unsupervised optical flow

    Rico Jonschkowski, Austin Stone, Jonathan T Barron, Ariel Gordon, Kurt Konolige, and Anelia Angelova. What mat- ters in unsupervised optical flow. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23– 28, 2020, Proceedings, Part II 16, pages 557–572. Springer,

    work page 2020

  39. [39]

    Repurpos- ing diffusion-based image generators for monocular depth estimation

    Bingxin Ke, Anton Obukhov, Shengyu Huang, Nando Met- zger, Rodrigo Caye Daudt, and Konrad Schindler. Repurpos- ing diffusion-based image generators for monocular depth estimation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9492– 9502, 2024. 2, 3

    work page 2024

  40. [40]

    The hci benchmark suite: Stereo and flow ground truth with uncertainties for urban autonomous driv- ing

    Daniel Kondermann, Rahul Nair, Katrin Honauer, Karsten Krispin, Jonas Andrulis, Alexander Brock, Burkhard Gusse- feld, Mohsen Rahimimoghaddam, Sabine Hofmann, Claus Brenner, et al. The hci benchmark suite: Stereo and flow ground truth with uncertainties for urban autonomous driv- ing. InProceedings of the IEEE Conference on Computer Vision and Pattern Rec...

    work page

  41. [41]

    Locally affine sparse-to-dense matching for motion and occlusion estimation

    Marius Leordeanu, Andrei Zanfir, and Cristian Sminchis- escu. Locally affine sparse-to-dense matching for motion and occlusion estimation. InProceedings of the IEEE Inter- national Conference on Computer Vision, pages 1721–1728,

    work page

  42. [42]

    Win-Win: Training High-Resolution Vision Transformers from Two Windows

    Vincent Leroy, Jerome Revaud, Thomas Lucas, and Philippe Weinzaepfel. Win-win: Training high-resolution vi- sion transformers from two windows.arXiv preprint arXiv:2310.00632, 2023. 2

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  43. [43]

    Fast guided global interpolation for depth and motion

    Yu Li, Dongbo Min, Minh N Do, and Jiangbo Lu. Fast guided global interpolation for depth and motion. InEu- ropean Conference on Computer Vision, pages 717–733. Springer, 2016. 1

    work page 2016

  44. [44]

    Megadepth: Learning single- view depth prediction from internet photos

    Zhengqi Li and Noah Snavely. Megadepth: Learning single- view depth prediction from internet photos. InProceed- ings of the IEEE conference on computer vision and pattern recognition, pages 2041–2050, 2018. 8

    work page 2041

  45. [45]

    Playing to vision foundation model’s strengths in stereo match- ing.IEEE Transactions on Intelligent Vehicles, 2024

    Chuang-Wei Liu, Qijun Chen, and Rui Fan. Playing to vision foundation model’s strengths in stereo match- ing.IEEE Transactions on Intelligent Vehicles, 2024. DOI:10.1109/TIV .2024.3467287. 3

    work page arXiv 2024

  46. [46]

    Learning by analogy: Reliable supervi- sion from transformations for unsupervised optical flow es- timation

    Liang Liu, Jiangning Zhang, Ruifei He, Yong Liu, Yabiao Wang, Ying Tai, Donghao Luo, Chengjie Wang, Jilin Li, and Feiyue Huang. Learning by analogy: Reliable supervi- sion from transformations for unsupervised optical flow es- timation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6489–6498,

  47. [47]

    Flow2stereo: Effective self-supervised learning of optical flow and stereo matching

    Pengpeng Liu, Irwin King, Michael R Lyu, and Jia Xu. Flow2stereo: Effective self-supervised learning of optical flow and stereo matching. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6648–6657, 2020. 2

    work page 2020

  48. [48]

    Unsupervised global and local ho- mography estimation with motion basis learning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(6):7885–7899, 2022

    Shuaicheng Liu, Yuhang Lu, Hai Jiang, Nianjin Ye, Chuan Wang, and Bing Zeng. Unsupervised global and local ho- mography estimation with motion basis learning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(6):7885–7899, 2022. 2 10

    work page 2022

  49. [49]

    Video frame inter- polation via optical flow estimation with image inpainting

    Xiaozhang Liu, Hui Liu, and Yuxiu Lin. Video frame inter- polation via optical flow estimation with image inpainting. International Journal of Intelligent Systems, 35(12):2087– 2102, 2020. 1

    work page 2087

  50. [50]

    Transflow: Trans- former as flow learner

    Yawen Lu, Qifan Wang, Siqi Ma, Tong Geng, Yingjie Victor Chen, Huaijin Chen, and Dongfang Liu. Transflow: Trans- former as flow learner. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18063–18073, 2023. 1

    work page 2023

  51. [51]

    Learning optical flow with kernel patch attention

    Ao Luo, Fan Yang, Xin Li, and Shuaicheng Liu. Learning optical flow with kernel patch attention. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8906–8915, 2022. 2

    work page 2022

  52. [52]

    A large dataset to train convolutional networks for disparity, optical flow, and scene flow estimation

    Nikolaus Mayer, Eddy Ilg, Philip Hausser, Philipp Fischer, Daniel Cremers, Alexey Dosovitskiy, and Thomas Brox. A large dataset to train convolutional networks for disparity, optical flow, and scene flow estimation. InProceedings of the IEEE conference on computer vision and pattern recog- nition, pages 4040–4048, 2016. 2, 6

    work page 2016

  53. [53]

    Spring: A high-resolution high- detail dataset and benchmark for scene flow, optical flow and stereo

    Lukas Mehl, Jenny Schmalfuss, Azin Jahedi, Yaroslava Nali- vayko, and Andr ´es Bruhn. Spring: A high-resolution high- detail dataset and benchmark for scene flow, optical flow and stereo. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4981– 4991, 2023. 2, 6, 7

    work page 2023

  54. [54]

    Object scene flow for au- tonomous vehicles

    Moritz Menze and Andreas Geiger. Object scene flow for au- tonomous vehicles. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 3061– 3070, 2015. 6, 7

    work page 2015

  55. [55]

    Recurrent partial kernel network for efficient optical flow estimation

    Henrique Morimitsu, Xiaobin Zhu, Xiangyang Ji, and Xu- Cheng Yin. Recurrent partial kernel network for efficient optical flow estimation. InProceedings of the AAAI Confer- ence on Artificial Intelligence, pages 4278–4286, 2024. 2, 6, 8

    work page 2024

  56. [56]

    Hello GPT-4o, 2024

    OpenAI. Hello GPT-4o, 2024. 2

    work page 2024

  57. [57]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. InInternational conference on machine learning, pages 8748–8763. PmLR, 2021. 2

    2021

  58. [58]

    Ren ´e Ranftl, Katrin Lasinger, David Hafner, Konrad Schindler, and Vladlen Koltun. Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer.IEEE transactions on pattern analysis and machine intelligence, 44(3):1623–1637, 2020. 3

    work page 2020

  59. [59]

    Vi- sion transformers for dense prediction.ArXiv preprint, 2021

    Ren ´e Ranftl, Alexey Bochkovskiy, and Vladlen Koltun. Vi- sion transformers for dense prediction.ArXiv preprint, 2021. 3, 4, 6

    work page 2021

  60. [60]

    Optical flow estima- tion using a spatial pyramid network

    Anurag Ranjan and Michael J Black. Optical flow estima- tion using a spatial pyramid network. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 4161–4170, 2017. 1, 2

    work page 2017

  61. [61]

    Epicflow: Edge-preserving interpolation of correspondences for optical flow

    Jerome Revaud, Philippe Weinzaepfel, Zaid Harchaoui, and Cordelia Schmid. Epicflow: Edge-preserving interpolation of correspondences for optical flow. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 1164–1172, 2015. 1

    work page 2015

  62. [62]

    Playing for benchmarks

    Stephan R Richter, Zeeshan Hayder, and Vladlen Koltun. Playing for benchmarks. InProceedings of the IEEE Inter- national Conference on Computer Vision, pages 2213–2222,

    work page

  63. [63]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022. 2

    2022

  64. [64]

    Multi-object discov- ery by low-dimensional object motion

    Sadra Safadoust and Fatma G ¨uney. Multi-object discov- ery by low-dimensional object motion. InProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. 2, 4

    work page 2023

  65. [65]

    The surprising effectiveness of diffusion models for optical flow and monocular depth estimation.Advances in Neural Information Processing Systems, 36, 2023

    Saurabh Saxena, Charles Herrmann, Junhwa Hur, Abhishek Kar, Mohammad Norouzi, Deqing Sun, and David J Fleet. The surprising effectiveness of diffusion models for optical flow and monocular depth estimation.Advances in Neural Information Processing Systems, 36, 2023. 2, 6, 7

    work page 2023

  66. [66]

    Videoflow: Exploiting temporal cues for multi-frame optical flow estimation

    Xiaoyu Shi, Zhaoyang Huang, Weikang Bian, Dasong Li, Manyuan Zhang, Ka Chun Cheung, Simon See, Hongwei Qin, Jifeng Dai, and Hongsheng Li. Videoflow: Exploiting temporal cues for multi-frame optical flow estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12469–12480, 2023. 2

    work page 2023

  67. [67]

    Flowformer++: Masked cost volume autoen- coding for pretraining optical flow estimation

    Xiaoyu Shi, Zhaoyang Huang, Dasong Li, Manyuan Zhang, Ka Chun Cheung, Simon See, Hongwei Qin, Jifeng Dai, and Hongsheng Li. Flowformer++: Masked cost volume autoen- coding for pretraining optical flow estimation. InProceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1599–1610, 2023. 1, 2, 8

    work page 2023

  68. [68]

    Craft: Cross- attentional flow transformer for robust optical flow

    Xiuchao Sui, Shaohua Li, Xue Geng, Yan Wu, Xinxing Xu, Yong Liu, Rick Goh, and Hongyuan Zhu. Craft: Cross- attentional flow transformer for robust optical flow. InPro- ceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition, pages 17602–17611, 2022. 1, 2, 6, 7, 8

    work page 2022

  69. [69]

    Secrets of optical flow estimation and their principles

    Deqing Sun, Stefan Roth, and Michael J Black. Secrets of optical flow estimation and their principles. In2010 IEEE computer society conference on computer vision and pattern recognition, pages 2432–2439. IEEE, 2010. 1

    work page 2010

  70. [70]

    Pwc-net: Cnns for optical flow using pyramid, warping, and cost volume

    Deqing Sun, Xiaodong Yang, Ming-Yu Liu, and Jan Kautz. Pwc-net: Cnns for optical flow using pyramid, warping, and cost volume. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 8934–8943,

  71. [71]

    Models matter, so does training: An empirical study of cnns for optical flow estimation.IEEE transactions on pattern analysis and machine intelligence, 42(6):1408–1423, 2019

    Deqing Sun, Xiaodong Yang, Ming-Yu Liu, and Jan Kautz. Models matter, so does training: An empirical study of cnns for optical flow estimation.IEEE transactions on pattern analysis and machine intelligence, 42(6):1408–1423, 2019. 1, 2

    work page 2019

  72. [72]

    Autoflow: Learning a better training set for optical flow

    Deqing Sun, Daniel Vlasic, Charles Herrmann, Varun Jampani, Michael Krainin, Huiwen Chang, Ramin Zabih, William T Freeman, and Ce Liu. Autoflow: Learning a better training set for optical flow. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10093–10102, 2021. 2, 7

    work page 2021

  73. [73]

    Disentan- 11 gling architecture and training for optical flow

    Deqing Sun, Charles Herrmann, Fitsum Reda, Michael Ru- binstein, David J Fleet, and William T Freeman. Disentan- 11 gling architecture and training for optical flow. InEuropean Conference on Computer Vision, pages 165–182. Springer,

    work page

  74. [74]

    Optical flow guided feature: A fast and robust motion representation for video action recognition

    Shuyang Sun, Zhanghui Kuang, Lu Sheng, Wanli Ouyang, and Wei Zhang. Optical flow guided feature: A fast and robust motion representation for video action recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 1390–1399, 2018. 1

    work page 2018

  75. [75]

    Skflow: Learning optical flow with super kernels

    Shangkun Sun, Yuanqi Chen, Yu Zhu, Guodong Guo, and Ge Li. Skflow: Learning optical flow with super kernels. Advances in Neural Information Processing Systems, 35: 11313–11326, 2022. 2, 6, 8

    work page 2022

  76. [76]

    Raft: Recurrent all-pairs field transforms for optical flow

    Zachary Teed and Jia Deng. Raft: Recurrent all-pairs field transforms for optical flow. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23– 28, 2020, Proceedings, Part II 16, pages 402–419. Springer,

    2020

  77. [77]

    Displacement-invariant matching cost learning for accurate optical flow estimation.Advances in Neural Information Processing Systems, 33, 2020

    Jianyuan Wang, Yiran Zhong, Yuchao Dai, Kaihao Zhang, Pan Ji, and Hongdong Li. Displacement-invariant matching cost learning for accurate optical flow estimation.Advances in Neural Information Processing Systems, 33, 2020. 1

    work page 2020

  78. [78]

    Tracking everything everywhere all at once

    Qianqian Wang, Yen-Yu Chang, Ruojin Cai, Zhengqi Li, Bharath Hariharan, Aleksander Holynski, and Noah Snavely. Tracking everything everywhere all at once. InProceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 19795–19806, 2023. 2

    work page 2023

  79. [79]

    Dust3r: Geometric 3d vi- sion made easy

    Shuzhe Wang, Vincent Leroy, Yohann Cabon, Boris Chidlovskii, and Jerome Revaud. Dust3r: Geometric 3d vi- sion made easy. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20697– 20709, 2024. 3

    2024

  80. [80]

    Tartanair: A dataset to push the limits of visual slam

    Wenshan Wang, Delong Zhu, Xiangwei Wang, Yaoyu Hu, Yuheng Qiu, Chen Wang, Yafei Hu, Ashish Kapoor, and Se- bastian Scherer. Tartanair: A dataset to push the limits of visual slam. In2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 4909–4916. IEEE, 2020. 2, 5, 6

    work page 2020

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.