Pith. sign in

REVIEW 5 major objections 6 minor 90 references

Benchmarking the Robustness of Optical Flow Estimation to Corruptions

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 This paper claims to build the first corruption robustness benchmark for optical flow estimation, covering 24 corruptions and 29 model variants on KITTI-FC and GoPro-FC.

desk verdict The first real corruption-robustness benchmark for optical flow, with a careful large-scale evaluation; the GoPro-FC results should be read as provisional because they rest on an unvalidated no-ground-truth proxy. read the letter →

arxiv 2411.14865 v1 pith:GNEDNDLD submitted 2024-11-22 eess.IV cs.CVcs.RO

Zhonghua Yi , Hao Shi , Qi Jiang , Yao Gao , Ze Wang , Yufan Zhang , Kailun Yang , Kaiwei Wang This is my paper · ORCID
classification eess.IVcs.CVcs.RO
keywords opticalflowestimationcorruptionrobustnessbenchmarktemporalcorruptionsKITTI-FCGoPro-FCmetricsout-of-domainevaluationunsupervised
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

The paper sets out to give optical flow estimation its first systematic way to measure how models behave when inputs are corrupted in ways common in real-world use, rather than only adversarially perturbed. It builds two benchmarks, KITTI-FC and GoPro-FC, spanning 24 corruptions, including seven temporal corruptions that exploit the two-frame nature of optical flow. Across 29 model variants it finds that a model's absolute robustness is largely determined by its clean-data accuracy, that corruptions destroying local information hurt more than mere visual degradation, and that unsupervised models are relatively stable. If the findings hold, robustness can be improved mainly by improving estimation performance, by using transformer-like and semantic features, and by preferring unsupervised training.

What carries the argument

The central mechanism is the benchmark itself, built from a corruption suite and paired evaluation protocols. The suite contains 24 corruptions in six classes, including seven temporal corruptions (Over Exposure, Under Exposure, Object Motion Blur, Snow, H.264 CRF, H.264 ABR, and Bit Error) that alter the two frames asymmetrically, alongside 17 single-image corruptions, with an upgraded PSF Blur generated from five simulated low-quality lenses. The benchmarks KITTI-FC and GoPro-FC respectively emphasize large and small displacements, and the metrics CRE, CREr, and RCRE quantify absolute and relative robustness, with RCRE usable when ground-truth flow is unavailable.

What would settle it

Corrupt only one frame of each pair, or use independently sampled per-frame spatter, frost, and fog overlays, then rerun RAFT, FlowFormer, and SAMFlow on KITTI-FC and compare the weather-robustness ranking with the shared-overlay setting; if the ranking changes materially, the benchmark's weather robustness numbers are an artifact of the identical-perturbation assumption.

Watch

Extended reading notes

Core claim

The paper's central claim is that optical flow robustness to common corruptions can and should be benchmarked systematically, and that the field has lacked such a benchmark. To support this, the authors introduce 24 corruptions split into six classes, with seven temporal corruptions specifically designed for image pairs, and construct KITTI-FC and GoPro-FC for autonomous driving and video editing scenarios, each with Out-Of-Domain and In-Domain settings. They propose Corruption Robustness Error (CRE), Corruption Robustness Error ratio (CREr), and Relative Corruption Robustness Error (RCRE) as metrics. Evaluating 29 model variants from 15 methods, they report ten observations, most notably that absolute robustness is heavily dependent on estimation performance, that corruptions which destroy local information are more serious than those which only reduce visual effects, and that unsupervised methods show strong relative robustness.

Load-bearing premise

The benchmark assumes that weather effects such as spatter, fog, and frost affect both frames of an image pair identically, applying the same corrupted rendering to both frames; if real weather artifacts differ subtly between frames, the measured weather robustness could overstate how well models handle true weather.

Editorial extensions

If this is right

  • Future optical flow evaluations can report CRE, CREr, and RCRE on KITTI-FC and GoPro-FC alongside EPE, making corruption robustness a standard comparison axis.
  • Improving clean-data estimation accuracy is a direct route to better absolute robustness, so progress on clean benchmarks should transfer to corrupted conditions.
  • Transformer-like architectures and semantic information help under large-displacement corruptions, but the benefit does not carry to small-displacement settings.
  • Unsupervised models are a practical robustness strategy: ARFlow achieves the best relative robustness with the smallest computing cost.
  • Fine-tuning to an in-domain target improves absolute robustness but lowers relative robustness, with weather corruptions becoming relatively more destabilizing.

Reading between the lines

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

  • A per-frame independent rendering of weather corruptions would be a stricter test; correlation-based models might match static spatter or frost patterns and appear more robust than a truly dynamic scene would allow.
  • Because CRE and RCRE are almost linearly related in the paper's own data, RCRE could stand in for CRE on any real video without ground truth, widening the benchmark to unlabeled footage.
  • The small-displacement failure of SAMFlow suggests semantic segmentation guidance is motion-scale dependent; finer or motion-aware masks could recover the benefit in GoPro-like settings.
  • Composite corruptions, which the paper lists as out of scope, are the natural next stress test and could be generated by composing the released per-corruption pipelines.
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

5 major / 6 minor

Summary. The paper introduces two corruption-robustness benchmarks for optical flow estimation, KITTI-FC and GoPro-FC, covering 24 corruptions in 6 classes, including 7 temporal corruptions, and proposes three robustness metrics: CRE, CREr, and RCRE. It evaluates 29 model variants from 15 optical flow methods under both Out-of-Domain and In-Domain settings for KITTI-FC and Out-of-Domain only for GoPro-FC, then derives 10 observations about model robustness, including the dependence of absolute robustness on clean estimation performance, the relative stability of unsupervised methods, the advantage of Transformer/SAM-based architectures in large-displacement settings, and the greater impact of corruptions that destroy local information. The benchmarks and code are promised for public release.

Significance. If the findings hold, this is a valuable and reusable evaluation suite for a relatively under-studied problem: corruption robustness of optical flow. The paper's strengths include the breadth of models evaluated (15 methods, 29 variants), the design of temporal corruptions specific to optical flow, the improved PSF blur simulation via designed low-quality lenses, the OOD/ID distinction, and the public release of the corrupted datasets. Several quantitative conclusions, however, are currently weaker than stated: the GoPro-FC rankings rest on a proxy metric validated only in a different displacement regime, the SAMFlow advantage is confounded with training budget, and Observation 1 is partly a consequence of the additive definition of CRE. The raw benchmark and tables are still a useful contribution, but the interpretive claims need tightening.

major comments (5)
  1. [5.3, Table 4, Eq. (3), Appendix D] Observations 9 and 10, including the claim that FlowFormer is the best robust model in small-displacement settings, rest entirely on RCRE, a no-ground-truth proxy. The validation in Appendix D and Fig. 9(d) is performed on KITTI-FC, a large-displacement benchmark, and reports only model-level linear correspondence between CRE and RCRE. This does not establish that RCRE-based model rankings match ground-truth-based robustness rankings in the small-displacement GoPro-FC regime, where a model can shift its prediction substantially without moving far from the ground truth, or shift little while crossing a motion boundary. The Limitations section (F.2) does not flag this. I ask the authors to validate RCRE on a small-displacement subset with ground truth (for example, by subsampling KITTI or Sintel pairs), report per-corruption rank correlations between RCRE and CRE or EPE-based robustness, and clearly qualify the GoPro-FC quantitative conclusions as proxy-based if such validation is not possible.
  2. [5.2, Observation 2; Tables 5 and 6] The paper attributes SAMFlow's superior robustness to the vision foundation model (SAM), but SAMFlow differs from FlowFormer in training schedule as well as architecture: SAMFlow is trained for 240k steps with batch size 3, while FlowFormer is trained for 120k with batch size 6. A longer training budget alone could explain part of the observed improvement, and the paper states that SAMFlow's gradient accumulation makes batch 3 equivalent to batch 6, but the step-count difference remains. Please retrain FlowFormer/FlowFormer++ with matched training steps (or ablate SAM within the SAMFlow pipeline) before claiming that the VFM provides the robustness benefit, or alternatively soften Observation 2 to reflect the confound.
  3. [4.1, Eq. (1); Observation 1] CRE is defined as EPE_c,s minus EPE_clean. Because EPE_clean appears as a subtractive term in the metric itself, a positive correlation between CRE and clean EPE is partly algebraic: even under a constant relative degradation, models with larger clean EPE will tend to have larger CRE. The paper's claim that 'absolute robustness depends heavily on estimation performance' therefore needs a decomposition that goes beyond the definitional component, for example by reporting partial correlations controlling for EPE_clean, or by showing that CREr (which removes this dependence) does not already account for the observed trend. As written, Observation 1 risks being tautological.
  4. [3 and A.3, Weather corruptions] For Spatter, Fog, and Frost, the same corrupted rendering is applied unchanged to both frames of the pair. Real weather artifacts are not static under camera and scene motion; an identical static overlay provides a zero-displacement matching clue that correlation-based optical flow estimators can exploit, potentially making weather robustness look better than it would be in practice and biasing model rankings. The paper's defense that these disturbances 'change slowly' is plausible for fog, but less so for spatter and frost, and it is not tested. I ask for a sensitivity analysis in which the overlay is independently sampled or slightly jittered between frames, with a report of whether the weather-class rankings and the overall observations change.
  5. [Tables 2-4; Observations 9-10] Several headline conclusions are based on small numerical differences with no error bars or significance tests. For example, on GoPro-FC the average RCRE values of FlowFormer (1.29), FlowFormer++ (1.35), and CRAFT (1.42) differ by less than 0.15, and many KITTI-FC corruption columns show similarly close values. Since the observations claim ordinal rankings among specific models, the authors should report at least bootstrap confidence intervals or paired significance tests for the main aggregate comparisons, so that readers can judge whether the reported ordering is meaningful.
minor comments (6)
  1. [1 vs. Abstract] The Introduction states that the authors found '7 interesting observations,' but the paper presents 10 observations. Please harmonize the count.
  2. [3 and A.3] There are several typos: 'Hight-Light' should be 'High-Light' in Section 3; 'we threat it unchanged' should be 'we treat it unchanged' in Appendix A.3; and the Table 2 header 'A VGEPE' should be 'AVG EPE'.
  3. [Appendix D, Fig. 9(d)] The claimed linear correspondence between CRE and RCRE is supported only by a scatter plot; please report the Pearson and Spearman correlation coefficients and, ideally, per-corruption values rather than only the model-level aggregate.
  4. [Table 3] The column 'Average Rank' is not clearly defined: it is not obvious whether the rank is averaged across corruptions, across metrics, or across something else. Please specify the averaging procedure.
  5. [Observation 5] The statement that CSFlow experiences 'an observed increase of up to 291% in CREr' is vague; please cite the specific corruption and condition that produce this maximum.
  6. [A.5, Object Motion Blur] The Object Motion Blur simulation relies on FLAVR video interpolation, which assumes that the interpolated high-frame-rate frames are physically accurate enough for blur accumulation. This assumption is not validated and should be acknowledged as a limitation of that specific corruption.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the benchmark construction and metric definitions are external to the claims, and the RCRE extrapolation is a validity concern, not a circular derivation.

full rationale

The paper's derivation chain is an empirical benchmarking study: corruptions are generated from external datasets (KITTI, GoPro) and external or self-contained simulation procedures, and the metrics CRE, CREr, and RCRE are explicitly defined by Eqs. (1)-(3). The observations are rankings and correlations computed from those definitions plus model outputs, not results derived from fitted parameters or from the definitions alone. Observation 1 (absolute robustness depends on estimation performance) is worth checking because CRE = EPE_corrupt - EPE_clean contains EPE_clean by definition; however, a positive correlation between CRE and EPE_clean is not forced by that definition (a model with large clean EPE could have small or negative CRE). The paper reports an empirical positive relationship and then introduces CREr = CRE / EPE_clean to remove the performance dependence, showing awareness of the definitional component rather than exploiting it. The GoPro-FC results rely on RCRE, a no-ground-truth proxy, and the paper validates RCRE against CRE on KITTI-FC (Fig. 9d) before transferring it to the small-displacement GoPro-FC setting. That transfer may be under-validated and is a legitimate correctness/validity concern, but it is not circular: RCRE is neither fitted to the GoPro-FC conclusions nor defined in terms of them. Self-citations to prior work (e.g., GSO for PSF lens generation, FLAVR for interpolation) are tool citations for benchmark construction, not load-bearing uniqueness arguments or ansatz-smuggling that reduce the paper's claims to the citations themselves. No step in the paper's reasoning exhibits the required equation-level or definition-level reduction to its own inputs, so the appropriate finding is no significant circularity.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The benchmark's conclusions rest on the realism of corruption simulation, the representativeness of two datasets, and the consistency of the evaluation protocol. No new physical entities are introduced.

free parameters (2)
  • Corruption severity parameters (hand-chosen per corruption) = e.g., JPEG quality [25,18,15,10,7]; noise std [0.08,...,0.38]; blur radii [3,...,10]
    These values define the five severity levels for each corruption. The paper explicitly states that parameters across different corruptions are not calibrated to a common real-world probability of occurrence, so cross-corruption comparisons of impact, such as Observation 8, rest on a hand-chosen severity scale rather than a physical or perceptual equivalence.
  • PSF lens RMS radii for blur simulation = 0.0296mm, 0.0832mm, 0.1102mm, 0.1588mm, 0.1939mm
    Five lens designs from the GSO framework are selected to represent low-quality optical systems. The choice of these specific lenses determines the PSF Blur difficulty, and the paper provides no independent evidence that they cover the range of real optical aberrations in deployed cameras.
assumptions (4)
  • domain assumption Spatter, Fog, and Frost are slow-changing and are applied identically to both frames of each image pair.
    Invoked in Section 3 (Weather corruptions) and Appendix A.3 to justify using the same corrupted rendering on both frames. If real weather patterns differ between frames, measured weather robustness may be optimistic.
  • domain assumption KITTI-2015 and GoPro are representative of autonomous driving and video editing optical flow applications.
    Section 4.2 selects these datasets for the two benchmarks. The generality of the 10 observations depends on this representativeness.
  • ad hoc to paper FLAVR video interpolation produces physically accurate high-frame-rate video for Object Motion Blur simulation.
    Section 4.2 (GoPro-FC construction) uses FLAVR for 4x interpolation to 960FPS and accumulates frames to synthesize object motion blur. Artifacts from the learned interpolator would make the corruption unrealistic.
  • domain assumption EPE on KITTI-FC is computed consistently with the official KITTI evaluation protocol, including the handling of invalid or occluded pixels.
    The paper does not specify the pixel mask used for EPE. Differences here would change CRE and CREr values and could affect the ranking of models.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Benchmarking the Robustness of Optical Flow Estimation to Corruptions." pith.science (2026). https://pith.science/paper/GNEDNDLD

@misc{pith2026241114865,
  author       = {Pith},
  title        = {Pith review of: Benchmarking the Robustness of Optical Flow Estimation to Corruptions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GNEDNDLD}},
  note         = {Machine review of arXiv:2411.14865}
}
read the original abstract

Optical flow estimation is extensively used in autonomous driving and video editing. While existing models demonstrate state-of-the-art performance across various benchmarks, the robustness of these methods has been infrequently investigated. Despite some research focusing on the robustness of optical flow models against adversarial attacks, there has been a lack of studies investigating their robustness to common corruptions. Taking into account the unique temporal characteristics of optical flow, we introduce 7 temporal corruptions specifically designed for benchmarking the robustness of optical flow models, in addition to 17 classical single-image corruptions, in which advanced PSF Blur simulation method is performed. Two robustness benchmarks, KITTI-FC and GoPro-FC, are subsequently established as the first corruption robustness benchmark for optical flow estimation, with Out-Of-Domain (OOD) and In-Domain (ID) settings to facilitate comprehensive studies. Robustness metrics, Corruption Robustness Error (CRE), Corruption Robustness Error ratio (CREr), and Relative Corruption Robustness Error (RCRE) are further introduced to quantify the optical flow estimation robustness. 29 model variants from 15 optical flow methods are evaluated, yielding 10 intriguing observations, such as 1) the absolute robustness of the model is heavily dependent on the estimation performance; 2) the corruptions that diminish local information are more serious than that reduce visual effects. We also give suggestions for the design and application of optical flow models. We anticipate that our benchmark will serve as a foundational resource for advancing research in robust optical flow estimation. The benchmarks and source code will be released at https://github.com/ZhonghuaYi/optical_flow_robustness_benchmark.

Figures

Figures reproduced from arXiv: 2411.14865 by the authors.

Figure 1
Figure 1. Benchmarks overview. (a) Clean image pairs are corrupted to construct optical flow robustness benchmarks KITTI-FC and GoPro-FC. (b) 29 model variants from 15 mainstream methods are evaluated through proposed robustness metrics in Out-Of-Domain (OOD) and In-Domain (ID) settings. SAMFlow-H [89] with Vision Foundation Model (VFM) [23], RAFT [70], and ARFlow [31] are taken as examples. (c) Comprehensive results and anal… view at source ↗
Figure 2
Figure 2. Effects of all the 24 corruptions under severity of 5. Corruptions are split into 6 classes. 7 temporal corruptions are in red boxes. The previous and next frames are displayed on the left and right sides of the image respectively for Over Exposure and Under Exposure for better visualization. Examples are from GoPro-FC. bustness, manipulating pixel [25] or patch information [58] to disrupt flow estimation. Recently,… view at source ↗
Figure 3
Figure 3. Calculating procedure of CREc,s and RCREc,s. RCREc,s is computed without using ground-truth optical flow. There are various proposed optical flow datasets, but most of them are synthetic [4, 9, 42, 43]. Real datasets [13, 24, 45] need large and heavy devices to record data and compute the optical flow ground-truth, thus all of them are capturing the scenes in city roads. For autonomous driving, we construct an optic… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The EPE, CRE, and CREr results of 12 optical flow models on OOD and ID benchmarks of KITTI-FC. CREr is represented by the size of the bubble and its value is indicated below the model name. Purple circles represent CNN-based models, green circles represent Transformer-…
Figure 5
Figure 5. Figure 5: CREr of representative models on different corrup￾tion classes. Weather corruptions heavily influence the supervised models from OOD to ID. best absolute robustness and relative robustness. Observation 4: The relative robustness of supervised meth￾ods drop in ID scenar…
Figure 6
Figure 6. Figure 6: SAM Segmentation results on (a) KITTI-FC and (b) GoPro-FC. Small motions in GoPro-FC result the pixel-level seg￾mentation misalignment not helpful for optical flow estimation. Saturate, Fog). For example, as shown in [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Optical flow amplitude statistic of KITTI-FC and GoPro-FC. GoPro-FC contains much more percent of small dis￾placement. benchmarks, as shown in [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 9
Figure 9. Figure 9: Relationship between clean EPE, corrupt EPE, CRE, and RCRE [PITH_FULL_IMAGE:figures/full_fig_p016_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

90 extracted references · 72 canonical work pages

  1. [1]

    CosPGD: an efficient white-box adversarial attack for pixel- wise prediction tasks

    Shashank Agnihotri, Steffen Jung, and Margret Keuper. CosPGD: an efficient white-box adversarial attack for pixel- wise prediction tasks. In ICML, 2024. 2

    2024

  2. [2]

    Anurag Arnab, Ondrej Miksik, and Philip H. S. Torr. On the robustness of semantic segmentation models to adversarial attacks. In CVPR, 2018. 2

    2018

  3. [3]

    Flow fields: Dense correspondence fields for highly accurate large displacement optical flow estimation

    Christian Bailer, Bertram Taetz, and Didier Stricker. Flow fields: Dense correspondence fields for highly accurate large displacement optical flow estimation. In ICCV, 2015. 2

    2015

  4. [4]

    Butler, Jonas Wulff, Garrett B

    Daniel J. Butler, Jonas Wulff, Garrett B. Stanley, and Michael J. Black. A naturalistic open source movie for opti- cal flow evaluation. In ECCV, 2012. 5, 14

    2012

  5. [5]

    Optical flow based visual potential field for autonomous driving

    Linda Capito, Umit Ozguner, and Keith Redmill. Optical flow based visual potential field for autonomous driving. In IV, 2020. 1, 4

    2020

  6. [6]

    Video controlnet: Towards temporally consistent synthetic-to-real video translation using conditional image diffusion models

    Ernie Chu, Shuo-Yen Lin, and Jun-Cheng Chen. Video controlnet: Towards temporally consistent synthetic-to-real video translation using conditional image diffusion models. arXiv preprint arXiv:2305.19193, 2023. 4

    arXiv 2023

  7. [7]

    MeDM: Mediating image diffusion models for video- to-video translation with temporal correspondence guidance

    Ernie Chu, Tzuhsuan Huang, Shuo-Yen Lin, and Jun-Cheng Chen. MeDM: Mediating image diffusion models for video- to-video translation with temporal correspondence guidance. In AAAI, 2024. 1, 4

    2024

  8. [8]

    Benchmarking robustness of 3D object detection to common corruptions in autonomous driving

    Yinpeng Dong, Caixin Kang, Jinlai Zhang, Zijian Zhu, Yikai Wang, Xiao Yang, Hang Su, Xingxing Wei, and Jun Zhu. Benchmarking robustness of 3D object detection to common corruptions in autonomous driving. In CVPR, 2023. 2, 3

    2023

Show all 90 references
  1. [9]

    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. InICCV,

  2. [10]

    Two-frame motion estimation based on polynomial expansion

    Gunnar Farneb ¨ack. Two-frame motion estimation based on polynomial expansion. In SCIA, 2003. 2, 5, 13

    2003

  3. [11]

    Review on panoramic imaging and its applications in scene understanding

    Shaohua Gao, Kailun Yang, Hao Shi, Kaiwei Wang, and Jian Bai. Review on panoramic imaging and its applications in scene understanding. IEEE Transactions on Instrumentation and Measurement, 2022. 4

    2022

  4. [12]

    Global search optics: Automatically ex- ploring optimal solutions to compact computational imaging systems

    Yao Gao, Qi Jiang, Shaohua Gao, Lei Sun, Kailun Yang, and Kaiwei Wang. Global search optics: Automatically ex- ploring optimal solutions to compact computational imaging systems. arXiv preprint arXiv:2404.19201, 2024. 4, 13

    2024 arXiv

  5. [13]

    Are we ready for autonomous driving? The KITTI vision benchmark suite

    Andreas Geiger, Philip Lenz, and Raquel Urtasun. Are we ready for autonomous driving? The KITTI vision benchmark suite. In CVPR, 2012. 5

    2012

  6. [14]

    Dietterich

    Dan Hendrycks and Thomas G. Dietterich. Benchmarking neural network robustness to common corruptions and per- turbations. In ICLR, 2019. 2, 3

    2019

  7. [15]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. In NeurIPS, 2020. 2

    2020

  8. [16]

    Berthold K. P. Horn and Brian G. Schunck. Determining optical flow. Artificial Intelligence, 1981. 2

    1981

  9. [17]

    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. In ECCV, 2022. 1, 2, 5, 14

    2022

  10. [18]

    FlowNet 2.0: Evolu- tion of optical flow estimation with deep networks

    Eddy Ilg, Nikolaus Mayer, Tonmoy Saikia, Margret Keuper, Alexey Dosovitskiy, and Thomas Brox. FlowNet 2.0: Evolu- tion of optical flow estimation with deep networks. InCVPR,

  11. [19]

    Computa- tional imaging for machine perception: Transferring seman- tic segmentation beyond aberrations

    Qi Jiang, Hao Shi, Shaohua Gao, Jiaming Zhang, Kailun Yang, Lei Sun, Huajian Ni, and Kaiwei Wang. Computa- tional imaging for machine perception: Transferring seman- tic segmentation beyond aberrations. IEEE Transactions on Computational Imaging, 2024. 2

    2024

  12. [20]

    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. In ICCV, 2021. 2, 5, 14

    2021

  13. [21]

    FLA VR: Flow-agnostic video representations for fast frame interpolation

    Tarun Kalluri, Deepak Pathak, Manmohan Chandraker, and Du Tran. FLA VR: Flow-agnostic video representations for fast frame interpolation. In WACV, 2023. 5, 13, 14

    2023

  14. [22]

    Benchmarking the robustness of semantic segmentation models

    Christoph Kamann and Carsten Rother. Benchmarking the robustness of semantic segmentation models. In CVPR,

  15. [23]

    Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross B

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chlo´e Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C. Berg, Wan-Yen Lo, Piotr Doll ´ar, and Ross B. Girshick. Segment anything. In ICCV, 2023. 1, 2, 5, 14

    2023

  16. [24]

    The HCI benchmark suite: Stereo and flow ground truth with uncertainties for urban au- tonomous driving

    Daniel Kondermann, Rahul Nair, Katrin Honauer, Karsten Krispin, Jonas Andrulis, Alexander Brock, Burkhard G¨ussefeld, Mohsen Rahimimoghaddam, Sabine Hofmann, Claus Brenner, and Bernd J¨ahne. The HCI benchmark suite: Stereo and flow ground truth with uncertainties for urban au-...

    2016

  17. [25]

    Consistent semantic attacks on optical flow

    Tom Koren, Lior Talker, Michael Dinerstein, and Ran Vitek. Consistent semantic attacks on optical flow. In ACCV, 2022. 3

    2022

  18. [26]

    Fast optical flow using dense inverse search

    Till Kroeger, Radu Timofte, Dengxin Dai, and Luc Van Gool. Fast optical flow using dense inverse search. InECCV, 2016. 1, 2, 5, 13

    2016

  19. [27]

    Adversarial attacks on video object segmen- tation with hard region discovery.IEEE Transactions on Cir- cuits and Systems for Video Technology, 2023

    Ping Li, Yu Zhang, Li Yuan, Jian Zhao, Xianghua Xu, and Xiaoqin Zhang. Adversarial attacks on video object segmen- tation with hard region discovery.IEEE Transactions on Cir- cuits and Systems for Video Technology, 2023. 2

    2023

  20. [28]

    Tan, and Loong-Fah Cheong

    Ruoteng Li, Robby T. Tan, and Loong-Fah Cheong. Robust optical flow in rainy scenes. In ECCV, 2018. 2

    2018

  21. [29]

    Tan, Loong Fah Cheong, Angelica I

    Ruoteng Li, Robby T. Tan, Loong Fah Cheong, Angelica I. Avil´es-Rivero, Qingnan Fan, and Carola Sch ¨onlieb. Rain- Flow: Optical flow under rain streaks and rain veiling effect. In ICCV, 2019. 2

    2019

  22. [30]

    FlowVid: Tam- ing imperfect optical flows for consistent video-to-video syn- thesis

    Feng Liang, Bichen Wu, Jialiang Wang, Licheng Yu, Kun- peng Li, Yinan Zhao, Ishan Misra, Jia-Bin Huang, Peizhao Zhang, Peter Vajda, and Diana Marculescu. FlowVid: Tam- ing imperfect optical flows for consistent video-to-video syn- thesis. In CVPR, 2024. 4

    2024

  23. [31]

    Learning by analogy: Reliable supervision from transformations for unsupervised optical flow estima- tion

    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 supervision from transformations for unsupervised optical flow estima- tion. In CVPR, 2020. 1, 2, 5, 14

    2020

  24. [32]

    SelF- low: Self-Supervised learning of optical flow

    Pengpeng Liu, Michael Lyu, Irwin King, and Jia Xu. SelF- low: Self-Supervised learning of optical flow. In CVPR, 2019

    2019

  25. [33]

    OIFlow: Occlusion-inpainting optical flow estimation by unsupervised learning

    Shuaicheng Liu, Kunming Luo, Nianjin Ye, Chuan Wang, Jue Wang, and Bing Zeng. OIFlow: Occlusion-inpainting optical flow estimation by unsupervised learning. IEEE Transactions on Image Processing, 2021. 2

    2021

  26. [34]

    Chen, and Dongfang Liu

    Yawen Lu, Qifan Wang, Siqi Ma, Tong Geng, Yingjie Vic- tor Chen, Huaijin G. Chen, and Dongfang Liu. TransFlow: Transformer as flow learner. In CVPR, 2023. 2

    2023

  27. [35]

    Bruce D. Lucas. An iterative technique of image registration and its application to stereo. In IJCAI, 1981. 1, 2

    1981

  28. [36]

    GAFlow: Incorporating gaussian attention into optical flow

    Ao Luo, Fan Yang, Xin Li, Lang Nie, Chunyu Lin, Haoqiang Fan, and Shuaicheng Liu. GAFlow: Incorporating gaussian attention into optical flow. In ICCV, 2023. 2

    2023

  29. [37]

    FlowDiffuser: Advancing optical flow esti- mation with diffusion models

    Ao Luo, Xin Li, Fan Yang, Jiangyu Liu, Haoqiang Fan, and Shuaicheng Liu. FlowDiffuser: Advancing optical flow esti- mation with diffusion models. In CVPR, 2024. 2, 5, 14

    2024

  30. [38]

    UPFlow: Upsampling pyramid for unsupervised optical flow learning

    Kunming Luo, Chuan Wang, Shuaicheng Liu, Haoqiang Fan, Jue Wang, and Jian Sun. UPFlow: Upsampling pyramid for unsupervised optical flow learning. In CVPR, 2021. 2, 5, 14

    2021

  31. [39]

    PoseBench: Benchmarking the robustness of pose estimation models under corruptions

    Sihan Ma, Jing Zhang, Qiong Cao, and Dacheng Tao. PoseBench: Benchmarking the robustness of pose estimation models under corruptions. arXiv preprint arXiv:2406.14367,

    arXiv

  32. [40]

    BrightFlow: Brightness-change-aware un- supervised learning of optical flow

    R ´emi Marsal, Florian Chabot, Ang ´elique Loesch, and Hichem Sahbi. BrightFlow: Brightness-change-aware un- supervised learning of optical flow. In WACV, 2023. 2, 5, 14

    2023

  33. [41]

    Alexander Mathis, Thomas Biasi, Steffen Schneider, Mert Y¨uksekg¨on¨ul, Byron Rogers, Matthias Bethge, and Macken- zie W. Mathis. Pretraining boosts out-of-domain robustness for pose estimation. In WACV, 2021. 2

    2021

  34. [42]

    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. In CVPR, 2016. 5, 14

    2016

  35. [43]

    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. In CVPR, 2023. 5

    2023

  36. [44]

    UnFlow: Un- supervised learning of optical flow with a bidirectional cen- sus loss

    Simon Meister, Junhwa Hur, and Stefan Roth. UnFlow: Un- supervised learning of optical flow with a bidirectional cen- sus loss. In AAAI, 2018. 2

    2018

  37. [45]

    Object scene flow for autonomous vehicles

    Moritz Menze and Andreas Geiger. Object scene flow for autonomous vehicles. In CVPR, 2015. 5, 14

    2015

  38. [46]

    Ecker, Matthias Bethge, and Wieland Brendel

    Claudio Michaelis, Benjamin Mitzkus, Robert Geirhos, Evgenia Rusak, Oliver Bringmann, Alexander S. Ecker, Matthias Bethge, and Wieland Brendel. Benchmarking ro- bustness in object detection: Autonomous driving when win- ter is coming. arXiv preprint arXiv:1907.07484, 2019. 2, 4, 12

    1907 arXiv

  39. [47]

    Meta- learning for adaptation of deep optical flow networks

    Chaerin Min, Taehyun Kim, and Jongwoo Lim. Meta- learning for adaptation of deep optical flow networks. In WACV, 2023. 2

    2023

  40. [48]

    Clas- sification robustness to common optical aberrations

    Patrick M ¨uller, Alexander Braun, and Margret Keuper. Clas- sification robustness to common optical aberrations. In IC- CVW, 2023. 2, 4

    2023

  41. [49]

    Deep multi-scale convolutional neural network for dynamic scene deblurring

    Seungjun Nah, Tae Hyun Kim, and Kyoung Mu Lee. Deep multi-scale convolutional neural network for dynamic scene deblurring. In CVPR, 2017. 5, 14

    2017

  42. [50]

    Robustness of SAM: Segment anything under corruptions and beyond

    Yu Qiao, Chaoning Zhang, Taegoo Kang, Donghun Kim, Chenshuang Zhang, and Choong Seon Hong. Robustness of SAM: Segment anything under corruptions and beyond. arXiv preprint arXiv:2306.07713, 2023. 2

    2023 arXiv

  43. [51]

    Anurag Ranjan, Joel Janai, Andreas Geiger, and Michael J. Black. Attacking optical flow. In ICCV, 2019. 2

    2019

  44. [52]

    Unsupervised deep learning for optical flow estimation

    Zhe Ren, Junchi Yan, Bingbing Ni, Bin Liu, Xiaokang Yang, and Hongyuan Zha. Unsupervised deep learning for optical flow estimation. In AAAI, 2017. 2

    2017

  45. [53]

    EpicFlow: Edge-preserving interpolation of correspondences for optical flow

    J ´erˆome Revaud, Philippe Weinzaepfel, Za¨ıd Harchaoui, and Cordelia Schmid. EpicFlow: Edge-preserving interpolation of correspondences for optical flow. In CVPR, 2015. 1, 2

    2015

  46. [54]

    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. In NeurIPS, 2023. 2

    2023

  47. [55]

    Robustness analysis of video- language models against visual and language perturbations

    Madeline Schiappa, Shruti Vyas, Hamid Palangi, Yogesh Rawat, and Vibhav Vineet. Robustness analysis of video- language models against visual and language perturbations. NeurIPS, 2022. 3

    2022

  48. [56]

    A perturbation-constrained adversarial attack for evaluating the robustness of optical flow

    Jenny Schmalfuss, Philipp Scholze, and Andr ´es Bruhn. A perturbation-constrained adversarial attack for evaluating the robustness of optical flow. In ECCV, 2022. 2

    2022

  49. [57]

    Distract- ing downpour: Adversarial weather attacks for motion esti- mation

    Jenny Schmalfuss, Lukas Mehl, and Andr ´es Bruhn. Distract- ing downpour: Adversarial weather attacks for motion esti- mation. In ICCV, 2023. 2, 3

    2023

  50. [58]

    To- wards understanding adversarial robustness of optical flow networks

    Simon Schrodi, Tonmoy Saikia, and Thomas Brox. To- wards understanding adversarial robustness of optical flow networks. In CVPR, 2022. 3

    2022

  51. [59]

    CSFlow: Learning optical flow via cross strip correlation for autonomous driving

    Hao Shi, Yifan Zhou, Kailun Yang, Xiaoting Yin, and Kai- wei Wang. CSFlow: Learning optical flow via cross strip correlation for autonomous driving. In IV, 2022. 2, 5, 14

    2022

  52. [60]

    PanoFlow: Learning 360° optical flow for surrounding temporal understanding

    Hao Shi, Yifan Zhou, Kailun Yang, Xiaoting Yin, Ze Wang, Yaozu Ye, Zhe Yin, Shi Meng, Peng Li, and Kaiwei Wang. PanoFlow: Learning 360° optical flow for surrounding temporal understanding. IEEE Transactions on Intelligent Transportation Systems, 2023. 2

    2023

  53. [61]

    Beyond the field-of-view: Enhanc- ing scene visibility and perception with clip-recurrent trans- former

    Hao Shi, Qi Jiang, Kailun Yang, Xiaoting Yin, Huajian Ni, and Kaiwei Wang. Beyond the field-of-view: Enhanc- ing scene visibility and perception with clip-recurrent trans- former. IEEE Transactions on Intelligent Vehicles, 2024. 4, 15

    2024

  54. [62]

    Benchmarking out-of- distribution detection in visual question answering

    Xiangxi Shi and Stefan Lee. Benchmarking out-of- distribution detection in visual question answering. In WACV, 2024. 2

    2024

  55. [63]

    FlowFormer++: Masked cost volume au- toencoding 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 au- toencoding for pretraining optical flow estimation. In CVPR,

  56. [64]

    Denoising diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502, 2020. 2

    2010 arXiv

  57. [65]

    SMURF: Self-teaching multi-frame unsupervised RAFT with full-image warping

    Austin Stone, Daniel Maurer, Alper Ayvaci, Anelia An- gelova, and Rico Jonschkowski. SMURF: Self-teaching multi-frame unsupervised RAFT with full-image warping. In CVPR, 2021. 2

    2021

  58. [66]

    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. InCVPR,

  59. [67]

    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. In CVPR, 2018. 1, 2

    2018

  60. [68]

    Event-based fusion for motion deblurring with cross- modal attention

    Lei Sun, Christos Sakaridis, Jingyun Liang, Qi Jiang, Kailun Yang, Peng Sun, Yaozu Ye, Kaiwei Wang, and Luc Van Gool. Event-based fusion for motion deblurring with cross- modal attention. In ECCV, 2022. 5

    2022

  61. [69]

    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. In NeurIPS, 2022. 2, 5, 14

    2022

  62. [70]

    RAFT: Recurrent all-pairs field transforms for optical flow

    Zachary Teed and Jia Deng. RAFT: Recurrent all-pairs field transforms for optical flow. In ECCV, 2020. 1, 2, 5, 13, 14

    2020

  63. [71]

    End-to-end interactive prediction and planning with op- tical flow distillation for autonomous driving

    Hengli Wang, Peide Cai, Rui Fan, Yuxiang Sun, and Ming Liu. End-to-end interactive prediction and planning with op- tical flow distillation for autonomous driving. In CVPRW,

  64. [72]

    Predbench: Benchmarking spatio-temporal prediction across diverse disciplines

    ZiDong Wang, Zeyu Lu, Di Huang, Tong He, Xihui Liu, Wanli Ouyang, and Lei Bai. Predbench: Benchmarking spatio-temporal prediction across diverse disciplines. arXiv preprint arXiv:2407.08418, 2024. 3

    2024 arXiv

  65. [73]

    GMFlow: Learning optical flow via global matching

    Haofei Xu, Jing Zhang, Jianfei Cai, Hamid Rezatofighi, and Dacheng Tao. GMFlow: Learning optical flow via global matching. In CVPR, 2022. 2

    2022

  66. [74]

    Adversarial attack against urban scene segmentation for autonomous vehicles

    Xing Xu, Jingran Zhang, Yujie Li, Yichuan Wang, Yang Yang, and Heng Tao Shen. Adversarial attack against urban scene segmentation for autonomous vehicles. IEEE Trans- actions on Industrial Informatics, 2021. 2

    2021

  67. [75]

    Wending Yan, Aashish Sharma, and Robby T. Tan. Op- tical flow estimation in dense foggy scenes with domain- adaptive networks. IEEE Transactions on Artificial Intelli- gence, 2023. 2

    2023

  68. [76]

    Benchmarking the robustness of LiDAR semantic segmentation models

    Xu Yan, Chaoda Zheng, Ying Xue, Zhen Li, Shuguang Cui, and Dengxin Dai. Benchmarking the robustness of LiDAR semantic segmentation models. International Journal of Computer Vision, 2024. 2

    2024

  69. [77]

    Rerender a video: Zero-shot text-guided video-to-video translation

    Shuai Yang, Yifan Zhou, Ziwei Liu, and Chen Change Loy. Rerender a video: Zero-shot text-guided video-to-video translation. In SIGGRAPH Asia, 2023. 1, 4

    2023

  70. [78]

    Benchmarking the robustness of spatial-temporal models against corruptions

    Chenyu Yi, Siyuan Yang, Haoliang Li, Yap-peng Tan, and Alex Kot. Benchmarking the robustness of spatial-temporal models against corruptions. In NeurIPS, 2021. 2, 3, 13

    2021

  71. [79]

    FocusFlow: Boosting key-points optical flow estimation for autonomous driving

    Zhonghua Yi, Hao Shi, Kailun Yang, Qi Jiang, Yaozu Ye, Ze Wang, Huajian Ni, and Kaiwei Wang. FocusFlow: Boosting key-points optical flow estimation for autonomous driving. IEEE Transactions on Intelligent Vehicles, 2023. 2

    2023

  72. [80]

    Benchmarking segmentation models with mask- preserved attribute editing

    Zijin Yin, Kongming Liang, Bing Li, Zhanyu Ma, and Jun Guo. Benchmarking segmentation models with mask- preserved attribute editing. In CVPR, 2024. 2

    2024

  73. [81]

    Optical flow domain adaptation via target style transfer

    Jeongbeen Yoon, Sanghyun Kim, Suha Kwak, and Minsu Cho. Optical flow domain adaptation via target style transfer. In WACV, 2024. 2

    2024

  74. [82]

    UnSAMFlow: Un- supervised optical flow guided by segment anything model

    Shuai Yuan, Lei Luo, Zhuo Hui, Can Pu, Xiaoyu Xiang, Rakesh Ranjan, and Denis Demandolx. UnSAMFlow: Un- supervised optical flow guided by segment anything model. In CVPR, 2024. 2

    2024

  75. [83]

    Benchmarking the robust- ness of temporal action detection models against temporal corruptions

    Runhao Zeng, Xiaoyong Chen, Jiaming Liang, Huisi Wu, Guangzhong Cao, and Yong Guo. Benchmarking the robust- ness of temporal action detection models against temporal corruptions. In CVPR, 2024. 2

    2024

  76. [84]

    Woodford, Victor Prisacariu, and Philip H

    Feihu Zhang, Oliver J. Woodford, Victor Prisacariu, and Philip H. S. Torr. Separable flow: Learning motion cost vol- umes for optical flow estimation. In ICCV, 2021. 2

    2021

  77. [85]

    Delivering arbitrary-modal semantic segmen- tation

    Jiaming Zhang, Ruiping Liu, Hao Shi, Kailun Yang, Simon Reiß, Kunyu Peng, Haodong Fu, Kaiwei Wang, and Rainer Stiefelhagen. Delivering arbitrary-modal semantic segmen- tation. In CVPR, 2023. 2

    2023

  78. [86]

    Shiyu Zhao, Long Zhao, Zhixing Zhang, Enyu Zhou, and Dimitris N. Metaxas. Global matching with overlapping at- tention for optical flow estimation. In CVPR, 2022. 2, 5, 14

    2022

  79. [87]

    Optical flow in the dark

    Yinqiang Zheng, Mingfang Zhang, and Feng Lu. Optical flow in the dark. In CVPR, 2020. 2

    2020

  80. [88]

    Unsu- pervised cumulative domain adaptation for foggy scene op- tical flow

    Hanyu Zhou, Yi Chang, Wending Yan, and Luxin Yan. Unsu- pervised cumulative domain adaptation for foggy scene op- tical flow. In CVPR, 2023. 2

    2023

  81. [89]

    SAMFlow: Eliminating any fragmentation in optical flow with segment anything model

    Shili Zhou, Ruian He, Weimin Tan, and Bo Yan. SAMFlow: Eliminating any fragmentation in optical flow with segment anything model. In AAAI, 2024. 1, 2, 5, 14

    2024

  82. [90]

    Understanding the robustness of 3D object detection with bird’s-eye-view representations in autonomous driving

    Zijian Zhu, Yichi Zhang, Hai Chen, Yinpeng Dong, Shu Zhao, Wenbo Ding, Jiachen Zhong, and Shibao Zheng. Understanding the robustness of 3D object detection with bird’s-eye-view representations in autonomous driving. In CVPR, 2023. 2, 4 A. Corruption Details A.1. Digital corrup...

    2023

Pith tools

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