Pith. sign in

REVIEW 3 major objections 5 minor 72 references

CompactFlowNet: Efficient Real-time Optical Flow Estimation on Mobile Devices

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read CompactFlowNet runs in 40 ms per 512x512 frame on an iPhone 8 while keeping accuracy on par with lightweight baselines.

desk verdict A useful mobile optical flow system with a real speed/memory win, but the abstract's accuracy claim is contradicted by the paper's own test tables. read the letter →

arxiv 2412.13273 v1 pith:7AC4WZMS submitted 2024-12-17 cs.CV

Andrei Znobishchev , Valerii Filev , Oleg Kudashev , Nikita Orlov , Humphrey Shi This is my paper · ORCID
classification cs.CV
keywords opticalflowestimationmobileinferencereal-timevideoprocessingCompactNetknowledgedistillationdepthwiseseparableconvolutionsPWC-Net
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

CompactFlowNet is an optical flow network built to run on phones. Optical flow predicts the per-pixel motion between two frames, and it is the backbone of video stabilization, tracking, action recognition, and video generation, so a fast on-device version makes those tasks possible without uploading video to a server. The paper's central claim is that CompactFlowNet is the first model to combine real-time mobile inference with accuracy competitive against lightweight baselines: 40 ms per 512x512 frame on an iPhone 8, 13 ms on an iPhone 14 Pro, and a peak memory footprint of 118 MB at that resolution. The design starts from a PWC-Net-style pyramid and reworks the expensive flow estimator, the refiner, and the backbone, then closes the accuracy gap with knowledge distillation from a heavier teacher.

What carries the argument

The load-bearing mechanism is architectural: the flow estimator, which dominates parameter count, FLOPs, and latency, is changed from a densely connected block to a sequentially connected block. Dense connections force every layer to receive concatenated maps from all earlier layers, keeping large tensors alive in memory and inflating both latency and the downstream refiner's input size; sequential connections remove that storage and compute burden. The other components are depthwise separable convolutions in the estimator and refiner, a shortened refiner with channels 128, 64, 32, and 2, and a MobileNetV3 backbone chosen by an on-device comparison of candidate backbones at 512x512 on an iPhone 8. Training adds a second mechanism: knowledge distillation from the heavier MobileNetV3-backed PWC-Net teacher, with loss $L = L_{\mathrm{sup}} + \gamma L_{\mathrm{dist}}$ at $\gamma = 0.1$, combined with an Autoflow pre-training dataset, extra augmentations, a OneCycle learning rate, and gradient clipping.

What would settle it

Reproduce the benchmark with the models converted under identical settings (same precision, same delegate, same iPhone OS) and check the 40 ms latency and 118 MB peak memory at 512x512 on an iPhone 8; if the margin over FastFlowNet disappears or the memory numbers grow past the reported values, the central claim fails.

Watch

Extended reading notes

Core claim

The paper argues that a PWC-Net-style coarse-to-fine optical flow model can be made mobile-friendly without sacrificing accuracy, by attacking the flow estimator block. Profiling on the iPhone 8 shows the estimator accounts for roughly 6.05 of the 8.75 million parameters and 131 of the 212 ms total latency of the original PWC-Net at 512x512. The authors replace the estimator's dense concatenation with a sequential connection pattern, adopt depthwise separable convolutions in the estimator and refiner, shrink the refiner from seven layers to four with channels 128, 64, 32, and 2, and swap the backbone for MobileNetV3 while keeping the full model under five million parameters. A heavy teacher model, the original architecture equipped with the MobileNetV3 backbone, is then used to distill CompactFlowNet, with a per-pixel L2 supervision loss plus a distillation term weighted at 0.1. The resulting model reaches 40 ms latency on an iPhone 8, 26 ms on an iPhone XR, 19 ms on an iPhone 12, and 13 ms on an iPhone 14 Pro at 512x512, with peak memory of 118 MB at that resolution; it also beats FastFlowNet and SpyNet on the Sintel test sets and runs faster than all compared lightweight models across three resolutions and four devices.

Load-bearing premise

The speed and memory comparisons assume that every model is converted to the mobile format with equally fair settings, so the measured differences on the phone reflect the architectures rather than conversion or precision artifacts.

Editorial extensions

If this is right

  • Tasks that depend on optical flow, such as stabilization, tracking, restoration, frame interpolation, and generative video editing, can be executed on-device without uploading footage.
  • The model runs at real-time speed (40 ms, 25 FPS) on an iPhone 8 and 13 ms on an iPhone 14 Pro at 512x512, with peak memory of 118 MB.
  • At Full HD (1080x1920) input it reaches nearly 10 FPS on an iPhone 14 Pro, and outperforms FastFlowNet and SpyNet on the Sintel test sets.
  • The under-10 MB float16 weight size and reduced memory footprint make the model compatible with the constraints the paper cites for mobile deployment.

Reading between the lines

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

  • The profiling table implies that dense feature concatenation, not the backbone, is the main latency and memory bottleneck in lightweight flow models; a direct activation-memory profiler could verify this on other PWC variants.
  • The accuracy gain from distillation (Sintel Final validation from 4.05 to 3.36, KITTI 2015 train from 2.14 to 1.91) suggests that a stronger teacher from a different model family might reduce the remaining error, although the paper reports trying such teachers without improvement.
  • The KITTI 2015 test f1-all of 14.64 is higher than FDFlowNet's 9.38, so the 'superior or comparable' claim should be read as benchmark-dependent; the speed and memory story does not depend on that comparison.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. CompactFlowNet is a proposed mobile-optimized optical flow network built on a PWC-Net-style architecture with a MobileNetV3 backbone, a sequentially connected flow estimator that removes dense concatenation, depthwise separable convolutions, and a reduced-depth flow refiner. The paper reports 40 ms average latency on an iPhone 8 at 512x512 input, lower peak memory than several lightweight competitors, and claims 'superior or comparable performance' to state-of-the-art lightweight models on KITTI and Sintel. Training uses a distillation pipeline with the authors' own heavy PWC-Net+MobileNetV3 model as teacher, and results are reported on Sintel and KITTI benchmarks as well as on-device latency and memory across four iPhone models.

Significance. If the reported speed and memory numbers are reproducible, the paper makes a useful engineering contribution: it demonstrates that a full optical flow network can run at interactive rates on a 2017-era phone, and it provides a cross-device latency and memory comparison for several lightweight methods. The paper ships no code or weights, however, and the accuracy claim is contradicted by the paper's own public test-set numbers, which show CompactFlowNet trailing FDFlowNet on every benchmark. The speed advantage may be real, but the overall claim as stated is overstated and the benchmarking methodology is not sufficiently specified to assess it.

major comments (3)
  1. [Abstract; Tables 4 and 5; Section 4.3] The abstract and Section 1 claim 'superior or comparable performance' to state-of-the-art lightweight models, but the paper's own public test-set results contradict this. On Sintel Clean test, CompactFlowNet reports 4.43 AEPE versus FDFlowNet's 3.71; on Sintel Final test, 5.55 versus 5.11; and on KITTI 2015 test (f1-all), 14.64 versus 9.38. CompactFlowNet is worse than FDFlowNet on every public test benchmark, with a relative KITTI gap exceeding 50%. The Discussion's statement (Section 4.3) that CompactFlowNet 'aligns closely with' FDFlowNet is inconsistent with these numbers, which are the conventional basis for benchmark comparison. The claim must be revised or re-scoped, for example to 'competitive accuracy at a fraction of the latency,' and the Discussion should honestly state the accuracy trade-off.
  2. [Section 4.2; Tables 6 and 7] The on-device speed and memory comparison is under-specified. The text says all models were 'implemented in PyTorch and converted to TFLite,' but it does not report the conversion tool versions, whether weights were quantized or cast to float16, which TFLite delegate (GPU or CPU) was used, or any per-model conversion settings. If competitor models convert less favorably, for example due to unsupported operators falling back to CPU, the claimed speed and memory advantage of CompactFlowNet could be an artifact of the conversion pipeline rather than of the architecture. The paper also gives no variance or error bars for the latency measurements, despite averaging over 100 runs. This is load-bearing for the paper's central speed claim.
  3. [Section 4.1; Section 3.1] The paper provides no code, pretrained weights, or complete training configuration, and it refers to 'specific augmentation parameters in Supplementary Materials' that are not present in the arXiv submission. The training pipeline includes many undocumented choices, such as the exact OneCycle schedule details, batch size, augmentation ranges, and distillation teacher configuration, which would be needed to reproduce the models. For a paper whose contribution is an architecture plus a training recipe, this lack of reproducibility evidence weakens the verification of both the accuracy and the speed claims.
minor comments (5)
  1. [Section 1] The abstract states 'the first real-time mobile neural network for optical flow prediction,' while the introduction uses 'To our knowledge, this is the first compact and memory-efficient model...' These are different claims; the hedging should be consistent, and the novelty claim should be supported by a more thorough survey of mobile deployment efforts.
  2. [Table 3] The parenthesized values in Table 3 are not sufficiently explained; the caption says they 'represent the results of networks that were trained on the same dataset,' but the reader has to infer which dataset is meant. Please clarify in the caption.
  3. [Table 6] The bold font for real-time compatible latency is helpful, but the threshold for real-time, for example 25 FPS or 40 ms, is not defined in the text or caption; please state it explicitly.
  4. [Section 3.2] The description of depthwise separable convolutions is standard, but it would help to state whether they replace all standard convolutions in the flow estimator and refiner, or only some; the text says 'adopted in both the flow estimator and flow refiner blocks' without specifying the extent.
  5. [Section 2] Reference [54] is cited for GPU latency and memory of RAFT and PWC-Net, but the numbers given in the text, such as 107 ms for RAFT, should be attributed explicitly to that paper to avoid ambiguity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; results are empirical measurements against external benchmarks.

full rationale

CompactFlowNet is an empirical systems paper: the accuracy claims are evaluated on the public KITTI and Sintel test sets against ground-truth flow, and the latency/memory claims are direct on-device measurements of PyTorch-to-TFLite conversions of all compared models. The teacher-student distillation (Section 3.4) uses the authors' heavy PWC-Net+MobileNetV3 as teacher, but the student's final evaluation is against ground-truth benchmarks (Tables 4-5), not against the teacher, and the supervision loss L_sup is computed on ground truth, so the architecture is not defined in terms of its own outputs. The backbone selection (Section 3.3) uses external benchmarks to choose among published backbones (MobileNetV3, ReXNet, HardCoReNAS), all cited as prior work; no load-bearing self-citation or imported uniqueness theorem appears. There is an apparent internal inconsistency between the abstract's 'superior or comparable performance' claim and the test-set rows of Tables 4-5 (e.g., CompactFlowNet's KITTI test f1-all 14.64 vs FDFlowNet 9.38), but this is a correctness/consistency issue, not circularity: the numbers are still measured against external data rather than derived from assumptions. Similarly, the on-device comparison assumes equal conversion quality across models (Section 4.2), but that is an experimental-condition concern, not a definitional reduction. No fitted parameter is relabeled as a prediction, and no claim is justified solely by a self-citation. Overall, the paper's central claims stand or fall on the reproducibility of its hardware measurements and benchmark results, not on circular reasoning.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claims are empirical, so the ledger is dominated by training and deployment choices rather than mathematical axioms. The two load-bearing assumptions are conversion fairness and hidden training details (missing supplement). No new physical or theoretical entities are introduced.

free parameters (6)
  • distillation loss weight gamma = 0.1
    Set by hand in Eq. 1 (L = Lsup + gamma * Ldist); no ablation over gamma reported.
  • flow refiner channel widths = 128, 64, 32, 2
    Chosen by hand in Section 3.2; no sensitivity analysis given.
  • flow refiner depth = 4 layers (from 7)
    Chosen by hand in Section 3.2; no ablation on layer count presented.
  • backbone choice = MobileNetV3 large (about 3.12M params)
    Selected empirically via Tables 2 and 3 as the best trade-off; this is a model-selection decision, not a derived optimum.
  • training iterations and learning schedule = 1.2M, 750k, 1.2M steps; OneCycle; weight decay 0 or 1e-5
    Adopted from prior flow-training work (Section 4.1); not ablated here.
  • augmentation parameters = not included in the arXiv submission
    Deferred to Supplementary Materials in Section 4.1, but the supplement is absent, so exact settings are unknowable from the preprint.
assumptions (5)
  • domain assumption PWC-Net is a suitable base model for mobile real-time optical flow.
    The entire design starts from profiling PWC-Net (Section 3). Starting from FastFlowNet or LiteFlowNet could have yielded different trade-offs.
  • domain assumption Conversion to TFLite is faithful and fair across all compared models.
    Latency and memory tables rely on this; no per-model conversion details, precision, or delegate are given (Sections 3.1, 4.2).
  • domain assumption 25 frames per second (about 40ms) is the right real-time threshold at 512x512 input.
    The paper treats a 40ms average latency as real-time; the threshold is application-dependent and not justified.
  • domain assumption Published benchmark scores of competitor methods are comparable to the authors' own runs.
    Train numbers are marked as same-dataset, but test numbers are official benchmark submissions with differences in training data and protocol that are not fully controlled.
  • domain assumption The teacher model is a good target for distillation.
    The heavy teacher is the authors' PWC-Net with MobileNetV3, not a published state-of-the-art flow model; its own accuracy is not reported or externally verified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CompactFlowNet: Efficient Real-time Optical Flow Estimation on Mobile Devices." pith.science (2026). https://pith.science/paper/7AC4WZMS

@misc{pith2026241213273,
  author       = {Pith},
  title        = {Pith review of: CompactFlowNet: Efficient Real-time Optical Flow Estimation on Mobile Devices},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7AC4WZMS}},
  note         = {Machine review of arXiv:2412.13273}
}
read the original abstract

We present CompactFlowNet, the first real-time mobile neural network for optical flow prediction, which involves determining the displacement of each pixel in an initial frame relative to the corresponding pixel in a subsequent frame. Optical flow serves as a fundamental building block for various video-related tasks, such as video restoration, motion estimation, video stabilization, object tracking, action recognition, and video generation. While current state-of-the-art methods prioritize accuracy, they often overlook constraints regarding speed and memory usage. Existing light models typically focus on reducing size but still exhibit high latency, compromise significantly on quality, or are optimized for high-performance GPUs, resulting in sub-optimal performance on mobile devices. This study aims to develop a mobile-optimized optical flow model by proposing a novel mobile device-compatible architecture, as well as enhancements to the training pipeline, which optimize the model for reduced weight, low memory utilization, and increased speed while maintaining minimal error. Our approach demonstrates superior or comparable performance to the state-of-the-art lightweight models on the challenging KITTI and Sintel benchmarks. Furthermore, it attains a significantly accelerated inference speed, thereby yielding real-time operational efficiency on the iPhone 8, while surpassing real-time performance levels on more advanced mobile devices.

Figures

Figures reproduced from arXiv: 2412.13273 by the authors.

Figure 1
Figure 1. Comparison of flow estimator structures: densely con [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Visualized results on Sintel Clean test set. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Visualized results on Sintel Final test set. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

72 extracted references · 59 canonical work pages

  1. [1]

    tensorflow lite

    Google llc. tensorflow lite. https://www.tensorflow. org/lite. Accessed: 2019-04-08. 4

    work page 2019

  2. [2]

    Mur- ray, Benoit Steiner, Paul Tucker, Vijay Vasudevan, Pete War- den, Martin Wicke, Yuan Yu, and Xiaoqiang Zheng

    Martin Abadi, Paul Barham, Jianmin Chen, Zhifeng Chen, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghe- mawat, Geoffrey Irving, Michael Isard, Manjunath Kudlur, Josh Levenberg, Rajat Monga, Sherry Moore, Derek G. Mur- ray, Benoit Steiner, Paul Tucker, Vijay Vasudevan, Pete War- den, Martin Wicke, Yuan Yu, and Xiaoqiang Zheng. Ten- sorflow: A system f...

    work page 2016

  3. [3]

    Video stabilization us- ing raft-based optical flow

    Rana Ashar, Burhan Sadiq, Hira Mohiuddin, Saniya Ashraf, Muhammad Imran, and Anayat Ullah. Video stabilization us- ing raft-based optical flow. In 2023 International Conference on Robotics and Automation in Industry (ICRAI), pages 1–5,

    work page 2023

  4. [4]

    Large displacement optical flow: descriptor matching in variational motion estimation

    Thomas Brox and Jitendra Malik. Large displacement optical flow: descriptor matching in variational motion estimation. IEEE transactions on pattern analysis and machine intelli- gence, 33(3):500–513, 2010. 1

    work page 2010

  5. [5]

    Mpi-sintel optical flow benchmark: Supplemental material

    D Butler, Jonas Wulff, G Stanley, and M Black. Mpi-sintel optical flow benchmark: Supplemental material. In MPI-IS- TR-006, MPI for Intelligent Systems (2012 . Citeseer, 2012. 5

    work page 2012

  6. [6]

    Basicvsr: The search for essential compo- nents in video super-resolution and beyond

    Kelvin CK Chan, Xintao Wang, Ke Yu, Chao Dong, and Chen Change Loy. Basicvsr: The search for essential compo- nents in video super-resolution and beyond. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4947–4956, 2021. 2, 3

    work page 2021

  7. [7]

    Improved optical flow for gesture-based human-robot interaction

    Jen-Yen Chang, Antonio Tejero-de Pablos, and Tatsuya Harada. Improved optical flow for gesture-based human-robot interaction. In 2019 International Conference on Robotics and Automation (ICRA), pages 7983–7989. IEEE, 2019. 1, 2

    work page 2019

  8. [8]

    Learning on-road visual control for self-driving vehicles with auxiliary tasks

    Yilun Chen, Palanisamy Praveen, Mudalige Priyantha, Kathe- rina Muelling, and John Dolan. Learning on-road visual control for self-driving vehicles with auxiliary tasks. In 2019 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 331–338. IEEE, 2019. 1

    work page 2019

Show all 72 references
  1. [9]

    Mfcflow: A motion feature compensated multi-frame recurrent network for optical flow estimation

    Yonghu Chen, Dongchen Zhu, Wenjun Shi, Guanghui Zhang, Tianyu Zhang, Xiaolin Zhang, and Jiamao Li. Mfcflow: A motion feature compensated multi-frame recurrent network for optical flow estimation. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Visi...

    2023

  2. [10]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009. 5

    2009

  3. [11]

    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. In Proceedings of the IEEE international conference on computer v...

    2015

  4. [12]

    End-to-end learning of motion representation for video understanding

    Lijie Fan, Wenbing Huang, Chuang Gan, Stefano Ermon, Boqing Gong, and Junzhou Huang. End-to-end learning of motion representation for video understanding. In Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 6016–6025, 2018. 2

    2018

  5. [13]

    Starflow: A spatiotemporal recurrent cell for lightweight multi-frame optical flow estimation

    Pierre Godet, Alexandre Boulch, Aur ´elien Plyer, and Guy Le Besnerais. Starflow: A spatiotemporal recurrent cell for lightweight multi-frame optical flow estimation. In 2020 25th International Conference on Pattern Recognition (ICPR), pages 2462–2469. IEEE, 2021. 2

    2020

  6. [14]

    Rethinking channel dimensions for efficient model design

    Dongyoon Han, Sangdoo Yun, Byeongho Heo, and YoungJoon Yoo. Rethinking channel dimensions for efficient model design. In Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition, pages 732–741,

  7. [15]

    Improving optical flow on a pyramid level

    Markus Hofinger, Samuel Rota Bul `o, Lorenzo Porzi, Arno Knapitsch, Thomas Pock, and Peter Kontschieder. Improving optical flow on a pyramid level. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23– 28, 2020, Proceedings, Part XXVIII, pages 770–786...

    2020

  8. [16]

    Determining optical flow

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

    1981

  9. [17]

    Searching for mo- bilenetv3

    Andrew Howard, Mark Sandler, Grace Chu, Liang-Chieh Chen, Bo Chen, Mingxing Tan, Weijun Wang, Yukun Zhu, Ruoming Pang, Vijay Vasudevan, et al. Searching for mo- bilenetv3. In Proceedings of the IEEE/CVF international conference on computer vision, pages 1314–1324, 2019. 4, 5

    2019

  10. [18]

    Mobilenets: Efficient convolu- tional neural networks for mobile vision applications

    Andrew G Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco An- dreetto, and Hartwig Adam. Mobilenets: Efficient convolu- tional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861, 2017. 2, 4

    2017 arXiv

  11. [19]

    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 Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, P...

    2022

  12. [20]

    Liteflownet3: Resolving correspondence ambiguity for more accurate optical flow estimation

    Tak-Wai Hui and Chen Change Loy. Liteflownet3: Resolving correspondence ambiguity for more accurate optical flow estimation. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XX 16, pages 169–184. Springer, 2020. 1

    2020

  13. [21]

    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. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 8981–8989, 2018. 1, 3

    2018

  14. [22]

    Iterative residual refinement for joint optical flow and occlusion estimation

    Junhwa Hur and Stefan Roth. Iterative residual refinement for joint optical flow and occlusion estimation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5754–5763, 2019. 1, 2

    2019

  15. [23]

    Ai benchmark: Running deep neural networks on android smartphones

    Andrey Ignatov, Radu Timofte, William Chou, Ke Wang, Max Wu, Tim Hartley, and Luc Van Gool. Ai benchmark: Running deep neural networks on android smartphones. In Proceedings of the European Conference on Computer Vision (ECCV) Workshops, pages 0–0, 2018. 2

    2018

  16. [24]

    Flownet 2.0: Evolution of optical flow estimation with deep networks

    Eddy Ilg, Nikolaus Mayer, Tonmoy Saikia, Margret Keu- per, Alexey Dosovitskiy, and Thomas Brox. Flownet 2.0: Evolution of optical flow estimation with deep networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2462–2470, 2017. 1, 2

    2017

  17. [25]

    Perceiver io: A general architecture for structured inputs & outputs

    Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Kop- pula, Daniel Zoran, Andrew Brock, Evan Shelhamer, et al. Perceiver io: A general architecture for structured inputs & outputs. arXiv preprint arXiv:2107.14795, 2021. 1, 3

    2021 arXiv

  18. [26]

    Unsupervised learning of multi-frame optical flow with occlusions

    Joel Janai, Fatma Guney, Anurag Ranjan, Michael Black, and Andreas Geiger. Unsupervised learning of multi-frame optical flow with occlusions. In Proceedings of the European conference on computer vision (ECCV), pages 690–706, 2018. 2

    2018

  19. [27]

    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 Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 9772– 9781, 2021. 1, 3

    2021

  20. [28]

    The hci benchmark suite: Stereo and flow ground truth with uncertainties for urban autonomous driving

    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 dr...

    2016

  21. [29]

    Fdflownet: Fast optical flow estimation using a deep lightweight network

    Lingtong Kong and Jie Yang. Fdflownet: Fast optical flow estimation using a deep lightweight network. In 2020 IEEE International Conference on Image Processing (ICIP), pages 1501–1505. IEEE, 2020. 1, 3

    2020

  22. [30]

    Fastflownet: A lightweight network for fast optical flow estimation

    Lingtong Kong, Chunhua Shen, and Jie Yang. Fastflownet: A lightweight network for fast optical flow estimation. In 2021 IEEE International Conference on Robotics and Automation (ICRA), pages 10310–10316. IEEE, 2021. 1, 2, 3

    2021

  23. [31]

    Fast optical flow using dense inverse search

    Till Kroeger, Radu Timofte, Dengxin Dai, and Luc Van Gool. Fast optical flow using dense inverse search. In Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part IV 14, pages 471–488. Springer, 2016. 1

    2016

  24. [32]

    On-device neural net inference with mobile gpus

    Juhyun Lee, Nikolay Chirkov, Ekaterina Ignasheva, Yury Pisarchyk, Mogan Shieh, Fabio Riccardi, Raman Sarokin, Andrei Kulik, and Matthias Grundmann. On-device neural net inference with mobile gpus. arXiv preprint arXiv:1907.01989,

    1907 arXiv

  25. [33]

    Centermask: Real-time anchor-free instance segmentation

    Youngwan Lee and Jongyoul Park. Centermask: Real-time anchor-free instance segmentation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recog- nition, pages 13906–13915, 2020. 4

    2020

  26. [34]

    An energy and gpu-computation efficient backbone network for real-time object detection

    Youngwan Lee, Joong-won Hwang, Sangrok Lee, Yuseok Bae, and Jongyoul Park. An energy and gpu-computation efficient backbone network for real-time object detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, pages 0–0, 2019. 4

    2019

  27. [35]

    Efficient- former: Vision transformers at mobilenet speed

    Yanyu Li, Geng Yuan, Yang Wen, Ju Hu, Georgios Evange- lidis, Sergey Tulyakov, Yanzhi Wang, and Jian Ren. Efficient- former: Vision transformers at mobilenet speed. Advances in Neural Information Processing Systems, 35:12934–12949,

  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 Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 9642– 9651, 2023. 3

    2023

  29. [37]

    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 Proceedings of the IEEE conference on computer vision and ...

    2016

  30. [38]

    Mobilevit: light- weight, general-purpose, and mobile-friendly vision trans- former

    Sachin Mehta and Mohammad Rastegari. Mobilevit: light- weight, general-purpose, and mobile-friendly vision trans- former. arXiv preprint arXiv:2110.02178, 2021. 5

    2021 arXiv

  31. [39]

    Object scene flow for autonomous vehicles

    Moritz Menze and Andreas Geiger. Object scene flow for autonomous vehicles. In Conference on Computer Vision and Pattern Recognition (CVPR), 2015. 5

    2015

  32. [40]

    Hardcore-nas: Hard constrained differentiable neural architec- ture search

    Niv Nayman, Yonathan Aflalo, Asaf Noy, and Lihi Zelnik. Hardcore-nas: Hard constrained differentiable neural architec- ture search. In International Conference on Machine Learn- ing, pages 7979–7990. PMLR, 2021. 5

    2021

  33. [41]

    Context-aware synthesis for video frame interpolation

    Simon Niklaus and Feng Liu. Context-aware synthesis for video frame interpolation. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 1701–1710, 2018. 2, 3

    2018

  34. [42]

    Pytorch: An imperative style, high-performance deep learning library

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Kopf, Edward Yang, Zachary DeVito, Martin Rai- son, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, L...

    2019

  35. [43]

    Optical flow estimation using a spatial pyramid network

    Anurag Ranjan and Michael J Black. Optical flow estimation using a spatial pyramid network. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4161–4170, 2017. 1, 3

    2017

  36. [44]

    Sudderth, and Jan Kautz

    Zhile Ren, Orazio Gallo, Deqing Sun, Ming-Hsuan Yang, Erik B. Sudderth, and Jan Kautz. A fusion approach for multi- frame optical flow estimation. In 2019 IEEE Winter Con- ference on Applications of Computer Vision (WACV), pages 2077–2086, 2019. 2

    2019

  37. [45]

    Richter, Zeeshan Hayder, and Vladlen Koltun

    Stephan R. Richter, Zeeshan Hayder, and Vladlen Koltun. Playing for benchmarks. In IEEE International Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22-29, 2017, pages 2232–2241, 2017. 5

    2017

  38. [46]

    Mobilenetv2: Inverted residuals and linear bottlenecks

    Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zh- moginov, and Liang-Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4510–4520, 2018. 4

    2018

  39. [47]

    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. arXiv preprint arXiv:2303.08340, 2023. 2, 6

    2023 arXiv

  40. [48]

    Flowformer++: Masked cost volume autoencoding 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 autoencoding for pretraining optical flow estimation. arXiv preprint arXiv:2303.01237, 2023. 1, 3, 6

    2023 arXiv

  41. [49]

    Fine-grained motion representation for template-free visual tracking

    Kai Shuang, Yuheng Huang, Yue Sun, Zhun Cai, and Hao Guo. Fine-grained motion representation for template-free visual tracking. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 671– 680, 2020. 1

    2020

  42. [50]

    Two-stream con- volutional networks for action recognition in videos

    Karen Simonyan and Andrew Zisserman. Two-stream con- volutional networks for action recognition in videos. In Pro- ceedings of the 27th International Conference on Neural In- formation Processing Systems - Volume 1 , page 568–576, Cambridge, MA, USA, 2014. MIT Press. 1

    2014

  43. [51]

    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 Proceedings of the IEEE conference on computer vision and pattern recognition, pages 8934–8943,

  44. [52]

    Models matter, so does training: An empirical study of cnns for optical flow estimation

    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. 5

    2019

  45. [53]

    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. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognitio...

    2021

  46. [54]

    Disentan- gling architecture and training for optical flow

    Deqing Sun, Charles Herrmann, Fitsum Reda, Michael Ru- binstein, David J Fleet, and William T Freeman. Disentan- gling architecture and training for optical flow. In Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Is- rael, October 23–27, 2022, Proceedings, Part...

    2022

  47. [55]

    Efficientnet: Rethinking model scaling for convolutional neural networks

    Mingxing Tan and Quoc Le. Efficientnet: Rethinking model scaling for convolutional neural networks. In International conference on machine learning, pages 6105–6114. PMLR,

  48. [56]

    Efficientnetv2: Smaller models and faster training

    Mingxing Tan and Quoc Le. Efficientnetv2: Smaller models and faster training. In International conference on machine learning, pages 10096–10106. PMLR, 2021. 4

    2021

  49. [57]

    Mixconv: Mixed depth- wise convolutional kernels

    Mingxing Tan and Quoc V Le. Mixconv: Mixed depth- wise convolutional kernels. arXiv preprint arXiv:1907.09595,

    1907 arXiv

  50. [58]

    Mnasnet: Platform-aware neural architecture search for mobile

    Mingxing Tan, Bo Chen, Ruoming Pang, Vijay Vasudevan, Mark Sandler, Andrew Howard, and Quoc V Le. Mnasnet: Platform-aware neural architecture search for mobile. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2820–2828, 2019. 4

    2019

  51. [59]

    Raft: Recurrent all-pairs field transforms for optical flow

    Zachary Teed and Jia Deng. Raft: Recurrent all-pairs field transforms for optical flow. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part II 16, pages 402–419. Springer, 2020. 1, 3, 6

    2020

  52. [60]

    Wadekar and Abhishek Chaurasia

    Shakti N. Wadekar and Abhishek Chaurasia. Mobilevitv3: Mobile-friendly vision transformer with simple and effective fusion of local, global and input features. 2022. 5

    2022

  53. [61]

    Correlation flow: robust optical flow using kernel cross- correlators

    Chen Wang, Tete Ji, Thien-Minh Nguyen, and Lihua Xie. Correlation flow: robust optical flow using kernel cross- correlators. In 2018 IEEE International Conference on Robotics and Automation (ICRA) , pages 836–841. IEEE,

    2018

  54. [62]

    Fbnet: Hardware-aware efficient con- vnet design via differentiable neural architecture search

    Bichen Wu, Xiaoliang Dai, Peizhao Zhang, Yanghan Wang, Fei Sun, Yiming Wu, Yuandong Tian, Peter Vajda, Yangqing Jia, and Kurt Keutzer. Fbnet: Hardware-aware efficient con- vnet design via differentiable neural architecture search. In Proceedings of the IEEE/CVF Conference on C...

    2019

  55. [63]

    High-resolution optical flow from 1d attention and correlation

    Haofei Xu, Jiaolong Yang, Jianfei Cai, Juyong Zhang, and Xin Tong. High-resolution optical flow from 1d attention and correlation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 10498–10507, 2021. 5

    2021

  56. [64]

    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 Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8121–8130,

  57. [65]

    Accurate optical flow via direct cost volume processing

    Jia Xu, Ren´e Ranftl, and Vladlen Koltun. Accurate optical flow via direct cost volume processing. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recogni- tion, pages 1289–1297, 2017. 1

    2017

  58. [66]

    Video enhancement with task-oriented flow

    Tianfan Xue, Baian Chen, Jiajun Wu, Donglai Wei, and William T Freeman. Video enhancement with task-oriented flow. International Journal of Computer Vision, 127:1106– 1125, 2019. 3

    2019

  59. [67]

    V olumetric correspon- dence networks for optical flow

    Gengshan Yang and Deva Ramanan. V olumetric correspon- dence networks for optical flow. Advances in neural informa- tion processing systems, 32, 2019. 1

    2019

  60. [68]

    Unsupervised motion representation enhanced network for action recogni- tion

    Xiaohang Yang, Lingtong Kong, and Jie Yang. Unsupervised motion representation enhanced network for action recogni- tion. In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 2445–2449. IEEE, 2021. 1

    2021

  61. [69]

    Learning video stabiliza- tion using optical flow

    Jiyang Yu and Ravi Ramamoorthi. Learning video stabiliza- tion using optical flow. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 8159–8167, 2020. 1, 2

    2020

  62. [70]

    A du- ality based approach for realtime tv-l 1 optical flow

    Christopher Zach, Thomas Pock, and Horst Bischof. A du- ality based approach for realtime tv-l 1 optical flow. In Pattern Recognition: 29th DAGM Symposium, Heidelberg, Germany, September 12-14, 2007. Proceedings 29 , pages 214–223. Springer, 2007. 1

    2007

  63. [71]

    Separable flow: Learning motion cost volumes for optical flow estimation

    Feihu Zhang, Oliver J Woodford, Victor Adrian Prisacariu, and Philip HS Torr. Separable flow: Learning motion cost volumes for optical flow estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 10807–10817, 2021. 5

    2021

  64. [72]

    Global matching with overlapping at- tention for optical flow estimation

    Shiyu Zhao, Long Zhao, Zhixing Zhang, Enyu Zhou, and Dimitris Metaxas. Global matching with overlapping at- tention for optical flow estimation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17592–17601, 2022. 1, 3

    2022

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.