{"url":"/task/optical-flow-estimation","name":"Optical Flow Estimation","slug":"optical-flow-estimation","description_markdown":"**Optical Flow Estimation** is a computer vision task that involves computing the motion of objects in an image or a video sequence. The goal of optical flow estimation is to determine the movement of pixels or features in the image, which can be used for various applications such as object tracking, motion analysis, and video compression.\r\n\r\nApproaches for optical flow estimation include correlation-based, block-matching, feature tracking, energy-based, and more recently gradient-based.\r\n\r\nFurther readings:\r\n\r\n- [Optical Flow Estimation](https://www.cs.toronto.edu/~fleet/research/Papers/flowChapter05.pdf)\r\n- [Performance of Optical Flow Techniques](https://www.cs.toronto.edu/~fleet/research/Papers/ijcv-94.pdf)\r\n\r\nDefinition source: [Devon: Deformable Volume Network for Learning Optical Flow ](https://arxiv.org/abs/1802.07351)\r\n\r\nImage credit: [Optical Flow Estimation](https://www.cs.toronto.edu/~fleet/research/Papers/flowChapter05.pdf)","categories":[{"name":"Computer Vision","url":"/area/computer-vision"}],"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","slug_source":"archive_url"},"counts":{"papers_tagged":2184,"papers_with_code":795,"benchmarks":10,"benchmark_tables_in_archive":10,"benchmark_tables_shown":10,"benchmark_tables_withheld_as_spam":0,"benchmark_definition":"a leaderboard table with at least one row; benchmark_tables_shown also counts the zero-row tables; benchmark_tables_in_archive adds the tables withheld as spam","datasets":35,"subtasks":1,"parent_tasks":0},"benchmarks":[{"leaderboard":"/sota/optical-flow-estimation-on-sintel-clean","slug":"optical-flow-estimation-on-sintel-clean","dataset":"Sintel-clean","dataset_url":"/dataset/mpi-sintel","rows_in_archive":29,"metrics":["Average End-Point Error"],"first_row_in_archive_order":{"model":"MEMFOF-L","paper_title":"MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation","paper_url":"/paper/memfof-high-resolution-training-for-memory","paper_date":"2025-06-29","arxiv_id":"2506.23151","code_links":[{"title":"msu-video-group/memfof","url":"https://github.com/msu-video-group/memfof"}],"syntology":{"n":13,"n_ran":9,"n_unverified":4,"n_pointer_only":0}}},{"leaderboard":"/sota/optical-flow-estimation-on-sintel-final","slug":"optical-flow-estimation-on-sintel-final","dataset":"Sintel-final","dataset_url":"/dataset/mpi-sintel","rows_in_archive":28,"metrics":["Average End-Point Error"],"first_row_in_archive_order":{"model":"MEMFOF-L","paper_title":"MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation","paper_url":"/paper/memfof-high-resolution-training-for-memory","paper_date":"2025-06-29","arxiv_id":"2506.23151","code_links":[{"title":"msu-video-group/memfof","url":"https://github.com/msu-video-group/memfof"}],"syntology":{"n":13,"n_ran":9,"n_unverified":4,"n_pointer_only":0}}},{"leaderboard":"/sota/optical-flow-estimation-on-kitti-2015-train","slug":"optical-flow-estimation-on-kitti-2015-train","dataset":"KITTI 2015 (train)","dataset_url":"/dataset/kitti","rows_in_archive":19,"metrics":["F1-all","EPE"],"first_row_in_archive_order":{"model":"MEMFOF","paper_title":"MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation","paper_url":"/paper/memfof-high-resolution-training-for-memory","paper_date":"2025-06-29","arxiv_id":"2506.23151","code_links":[{"title":"msu-video-group/memfof","url":"https://github.com/msu-video-group/memfof"}],"syntology":{"n":13,"n_ran":9,"n_unverified":4,"n_pointer_only":0}}},{"leaderboard":"/sota/optical-flow-estimation-on-kitti-2015","slug":"optical-flow-estimation-on-kitti-2015","dataset":"KITTI 2015","dataset_url":"/dataset/kitti","rows_in_archive":18,"metrics":["Fl-all","Average End-Point Error","Fl-fg"],"first_row_in_archive_order":{"model":"MEMFOF","paper_title":"MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation","paper_url":"/paper/memfof-high-resolution-training-for-memory","paper_date":"2025-06-29","arxiv_id":"2506.23151","code_links":[{"title":"msu-video-group/memfof","url":"https://github.com/msu-video-group/memfof"}],"syntology":{"n":13,"n_ran":9,"n_unverified":4,"n_pointer_only":0}}},{"leaderboard":"/sota/optical-flow-estimation-on-kitti-2012","slug":"optical-flow-estimation-on-kitti-2012","dataset":"KITTI 2012","dataset_url":"/dataset/kitti","rows_in_archive":12,"metrics":["Average End-Point Error","Out-Noc","Noc"],"first_row_in_archive_order":{"model":"CroCo-Flow","paper_title":"CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow","paper_url":"/paper/improved-cross-view-completion-pre-training","paper_date":"2022-11-18","arxiv_id":"2211.10408","code_links":[{"title":"naver/croco","url":"https://github.com/naver/croco"}],"syntology":null}},{"leaderboard":"/sota/optical-flow-estimation-on-spring","slug":"optical-flow-estimation-on-spring","dataset":"Spring","dataset_url":"/dataset/spring","rows_in_archive":11,"metrics":["1px total"],"first_row_in_archive_order":{"model":"MEMFOF","paper_title":"MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation","paper_url":"/paper/memfof-high-resolution-training-for-memory","paper_date":"2025-06-29","arxiv_id":"2506.23151","code_links":[{"title":"msu-video-group/memfof","url":"https://github.com/msu-video-group/memfof"}],"syntology":{"n":13,"n_ran":9,"n_unverified":4,"n_pointer_only":0}}},{"leaderboard":"/sota/optical-flow-estimation-on-sintel-clean-2","slug":"optical-flow-estimation-on-sintel-clean-2","dataset":"Sintel Clean unsupervised","dataset_url":"/dataset/mpi-sintel","rows_in_archive":5,"metrics":["Average End-Point Error"],"first_row_in_archive_order":{"model":"MDFlow","paper_title":"MDFlow: Unsupervised Optical Flow Learning by Reliable Mutual Knowledge Distillation","paper_url":"/paper/mdflow-unsupervised-optical-flow-learning-by","paper_date":"2022-11-11","arxiv_id":"2211.06018","code_links":[{"title":"ltkong218/mdflow","url":"https://github.com/ltkong218/mdflow"}],"syntology":{"n":2,"n_ran":2,"n_unverified":0,"n_pointer_only":1}}},{"leaderboard":"/sota/optical-flow-estimation-on-sintel-final-2","slug":"optical-flow-estimation-on-sintel-final-2","dataset":"Sintel Final unsupervised","dataset_url":"/dataset/mpi-sintel","rows_in_archive":5,"metrics":["Average End-Point Error"],"first_row_in_archive_order":{"model":"UpFlow","paper_title":"UPFlow: Upsampling Pyramid for Unsupervised Optical Flow Learning","paper_url":"/paper/upflow-upsampling-pyramid-for-unsupervised","paper_date":"2020-12-01","arxiv_id":"2012.00212","code_links":[{"title":"twhui/LiteFlowNet3","url":"https://github.com/twhui/LiteFlowNet3"},{"title":"coolbeam/UPFlow_pytorch","url":"https://github.com/coolbeam/UPFlow_pytorch"}],"syntology":{"n":1,"n_ran":1,"n_unverified":0,"n_pointer_only":0}}},{"leaderboard":"/sota/optical-flow-estimation-on-kitti-2015-2","slug":"optical-flow-estimation-on-kitti-2015-2","dataset":"KITTI 2015 unsupervised","dataset_url":"/dataset/kitti","rows_in_archive":4,"metrics":["Fl-all"],"first_row_in_archive_order":{"model":"MDFlow","paper_title":"MDFlow: Unsupervised Optical Flow Learning by Reliable Mutual Knowledge Distillation","paper_url":"/paper/mdflow-unsupervised-optical-flow-learning-by","paper_date":"2022-11-11","arxiv_id":"2211.06018","code_links":[{"title":"ltkong218/mdflow","url":"https://github.com/ltkong218/mdflow"}],"syntology":{"n":2,"n_ran":2,"n_unverified":0,"n_pointer_only":1}}},{"leaderboard":"/sota/optical-flow-estimation-on-kitti-2012-2","slug":"optical-flow-estimation-on-kitti-2012-2","dataset":"KITTI 2012 unsupervised","dataset_url":"/dataset/kitti","rows_in_archive":2,"metrics":["Average End-Point Error"],"first_row_in_archive_order":{"model":"UpFlow","paper_title":"UPFlow: Upsampling Pyramid for Unsupervised Optical Flow Learning","paper_url":"/paper/upflow-upsampling-pyramid-for-unsupervised","paper_date":"2020-12-01","arxiv_id":"2012.00212","code_links":[{"title":"twhui/LiteFlowNet3","url":"https://github.com/twhui/LiteFlowNet3"},{"title":"coolbeam/UPFlow_pytorch","url":"https://github.com/coolbeam/UPFlow_pytorch"}],"syntology":{"n":1,"n_ran":1,"n_unverified":0,"n_pointer_only":0}}}],"datasets":[{"url":"/dataset/kitti","name":"KITTI","full_name":"","num_papers_in_archive":3661},{"url":"/dataset/flyingthings3d","name":"FlyingThings3D","full_name":"","num_papers_in_archive":226},{"url":"/dataset/vimeo90k-1","name":"Vimeo90K","full_name":"","num_papers_in_archive":220},{"url":"/dataset/mpi-sintel","name":"MPI Sintel","full_name":"","num_papers_in_archive":198},{"url":"/dataset/megadepth","name":"MegaDepth","full_name":"","num_papers_in_archive":152},{"url":"/dataset/virtual-kitti","name":"Virtual KITTI","full_name":"","num_papers_in_archive":133},{"url":"/dataset/visdrone","name":"VisDrone","full_name":"","num_papers_in_archive":73},{"url":"/dataset/n-cars","name":"N-CARS","full_name":"","num_papers_in_archive":56},{"url":"/dataset/event-camera-dataset","name":"Event-Camera Dataset","full_name":"","num_papers_in_archive":51},{"url":"/dataset/spring","name":"Spring","full_name":"Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo","num_papers_in_archive":29},{"url":"/dataset/mvsec","name":"MVSEC","full_name":"Multi Vehicle Stereo Event Camera","num_papers_in_archive":28},{"url":"/dataset/dsec","name":"DSEC","full_name":"A Stereo Event Camera Dataset for Driving Scenarios","num_papers_in_archive":23},{"url":"/dataset/advio","name":"ADVIO","full_name":"","num_papers_in_archive":13},{"url":"/dataset/synwoodscape","name":"SynWoodScape","full_name":"Synthetic Surround-view Fisheye Camera Dataset for Autonomous Driving","num_papers_in_archive":13},{"url":"/dataset/yup","name":"YUP++","full_name":"YUP++ Dynamic Scenes dataset","num_papers_in_archive":13},{"url":"/dataset/slowflow","name":"SlowFlow","full_name":"","num_papers_in_archive":10},{"url":"/dataset/blackbird","name":"Blackbird","full_name":"","num_papers_in_archive":9},{"url":"/dataset/tum-gaid","name":"TUM-GAID","full_name":"TUM-GAID","num_papers_in_archive":9},{"url":"/dataset/samm-long-videos","name":"SAMM Long Videos","full_name":"","num_papers_in_archive":8},{"url":"/dataset/3dpeople-dataset","name":"3DPeople Dataset","full_name":"","num_papers_in_archive":6},{"url":"/dataset/omniflow","name":"OmniFlow","full_name":"","num_papers_in_archive":6},{"url":"/dataset/gof","name":"GOF","full_name":"Gyroscope Optical Flow","num_papers_in_archive":5},{"url":"/dataset/crowdflow","name":"CrowdFlow","full_name":"TUB CrowdFlow","num_papers_in_archive":4},{"url":"/dataset/eden","name":"EDEN","full_name":"","num_papers_in_archive":4},{"url":"/dataset/quva-repetition","name":"QUVA Repetition","full_name":"","num_papers_in_archive":4},{"url":"/dataset/vlog-dataset","name":"VLOG Dataset","full_name":"","num_papers_in_archive":4},{"url":"/dataset/hd1k","name":"HD1k","full_name":"","num_papers_in_archive":3},{"url":"/dataset/robotpush","name":"RobotPush","full_name":"RobotPush","num_papers_in_archive":3},{"url":"/dataset/spacenet-mvoi","name":"SpaceNet MVOI","full_name":"SpaceNet Multi-View Overhead Imagery Dataset","num_papers_in_archive":3},{"url":"/dataset/thirdtofirst","name":"ThirdToFirst","full_name":"","num_papers_in_archive":3},{"url":"/dataset/vbr","name":"VBR","full_name":"VBR: A Vision Benchmark in Rome","num_papers_in_archive":3},{"url":"/dataset/creative-flow-dataset","name":"Creative Flow+ Dataset","full_name":"","num_papers_in_archive":2},{"url":"/dataset/i2-2000fps","name":"I2-2000FPS","full_name":"","num_papers_in_archive":2},{"url":"/dataset/cloudcast","name":"CloudCast","full_name":"CloudCast: A Satellite-Based Dataset and Baseline for Forecasting Clouds","num_papers_in_archive":1},{"url":"/dataset/ekubric","name":"EKubric","full_name":"","num_papers_in_archive":1}],"subtasks":[{"url":"/task/video-stabilization","name":"Video Stabilization"}],"parent_tasks":[],"papers":{"order":"repositories listed in the archive (desc), then date (desc); the archive holds no stars","population":"papers tagged with this task that list at least one repository in the archive","shown":30,"of":795,"tagged_in_all":2184,"items":[{"url":"/paper/pwc-net-cnns-for-optical-flow-using-pyramid","title":"PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume","date":"2017-09-07","arxiv_id":"1709.02371","repositories_listed":21,"syntology":{"n":3,"n_ran":3,"n_unverified":0,"n_pointer_only":3}},{"url":"/paper/flownet-learning-optical-flow-with","title":"FlowNet: Learning Optical Flow with Convolutional Networks","date":"2015-04-26","arxiv_id":"1504.06852","repositories_listed":18,"syntology":{"n":20,"n_ran":1,"n_unverified":19,"n_pointer_only":0}},{"url":"/paper/raft-recurrent-all-pairs-field-transforms-for","title":"RAFT: Recurrent All-Pairs Field Transforms for Optical Flow","date":"2020-03-26","arxiv_id":"2003.12039","repositories_listed":17,"syntology":{"n":66,"n_ran":41,"n_unverified":25,"n_pointer_only":17}},{"url":"/paper/rife-real-time-intermediate-flow-estimation","title":"RIFE: Real-Time Intermediate Flow Estimation for Video Frame Interpolation","date":"2020-11-12","arxiv_id":"2011.06294","repositories_listed":13,"syntology":{"n":26,"n_ran":19,"n_unverified":7,"n_pointer_only":16}},{"url":"/paper/flownet-20-evolution-of-optical-flow","title":"FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks","date":"2016-12-06","arxiv_id":"1612.01925","repositories_listed":12,"syntology":{"n":21,"n_ran":2,"n_unverified":19,"n_pointer_only":3}},{"url":"/paper/perceiver-io-a-general-architecture-for","title":"Perceiver IO: A General Architecture for Structured Inputs & Outputs","date":"2021-07-30","arxiv_id":"2107.14795","repositories_listed":9,"syntology":{"n":11,"n_ran":7,"n_unverified":4,"n_pointer_only":0}},{"url":"/paper/optical-flow-estimation-using-a-spatial","title":"Optical Flow Estimation using a Spatial Pyramid Network","date":"2016-11-03","arxiv_id":"1611.00850","repositories_listed":8,"syntology":null},{"url":"/paper/two-stream-convolutional-networks-for-action","title":"Two-Stream Convolutional Networks for Action Recognition in Videos","date":"2014-06-09","arxiv_id":"1406.2199","repositories_listed":7,"syntology":{"n":7,"n_ran":2,"n_unverified":5,"n_pointer_only":2}},{"url":"/paper/semantic-flow-for-fast-and-accurate-scene","title":"Semantic Flow for Fast and Accurate Scene Parsing","date":"2020-02-24","arxiv_id":"2002.10120","repositories_listed":6,"syntology":{"n":8,"n_ran":1,"n_unverified":7,"n_pointer_only":1}},{"url":"/paper/video-frame-interpolation-via-adaptive","title":"Video Frame Interpolation via Adaptive Separable Convolution","date":"2017-08-05","arxiv_id":"1708.01692","repositories_listed":6,"syntology":null},{"url":"/paper/what-matters-in-unsupervised-optical-flow","title":"What Matters in Unsupervised Optical Flow","date":"2020-06-08","arxiv_id":"2006.04902","repositories_listed":5,"syntology":{"n":17,"n_ran":16,"n_unverified":1,"n_pointer_only":0}},{"url":"/paper/attention-guided-network-for-ghost-free-high","title":"Attention-guided Network for Ghost-free High Dynamic Range Imaging","date":"2019-04-23","arxiv_id":"1904.10293","repositories_listed":5,"syntology":null},{"url":"/paper/depth-aware-video-frame-interpolation","title":"Depth-Aware Video Frame Interpolation","date":"2019-04-01","arxiv_id":"1904.00830","repositories_listed":5,"syntology":{"n":15,"n_ran":10,"n_unverified":5,"n_pointer_only":0}},{"url":"/paper/representation-flow-for-action-recognition","title":"Representation Flow for Action Recognition","date":"2018-10-02","arxiv_id":"1810.01455","repositories_listed":5,"syntology":null},{"url":"/paper/super-slomo-high-quality-estimation-of","title":"Super SloMo: High Quality Estimation of Multiple Intermediate Frames for Video Interpolation","date":"2017-11-30","arxiv_id":"1712.00080","repositories_listed":5,"syntology":{"n":2,"n_ran":1,"n_unverified":1,"n_pointer_only":0}},{"url":"/paper/deep-multi-scale-video-prediction-beyond-mean","title":"Deep multi-scale video prediction beyond mean square error","date":"2015-11-17","arxiv_id":"1511.05440","repositories_listed":5,"syntology":{"n":10,"n_ran":2,"n_unverified":8,"n_pointer_only":0}},{"url":"/paper/bubbleml-a-multi-physics-dataset-and","title":"BubbleML: A Multi-Physics Dataset and Benchmarks for Machine Learning","date":"2023-07-27","arxiv_id":"2307.14623","repositories_listed":4,"syntology":{"n":10,"n_ran":7,"n_unverified":3,"n_pointer_only":10}},{"url":"/paper/gmflow-learning-optical-flow-via-global","title":"GMFlow: Learning Optical Flow via Global Matching","date":"2021-11-26","arxiv_id":"2111.13680","repositories_listed":4,"syntology":{"n":19,"n_ran":8,"n_unverified":11,"n_pointer_only":0}},{"url":"/paper/model-free-vehicle-tracking-and-state","title":"Model-free Vehicle Tracking and State Estimation in Point Cloud Sequences","date":"2021-03-10","arxiv_id":"2103.06028","repositories_listed":4,"syntology":null},{"url":"/paper/learning-accurate-dense-correspondences-and","title":"Learning Accurate Dense Correspondences and When to Trust Them","date":"2021-01-05","arxiv_id":"2101.01710","repositories_listed":4,"syntology":{"n":1,"n_ran":1,"n_unverified":0,"n_pointer_only":0}},{"url":"/paper/towards-better-generalization-joint-depth","title":"Towards Better Generalization: Joint Depth-Pose Learning without PoseNet","date":"2020-04-03","arxiv_id":"2004.01314","repositories_listed":4,"syntology":{"n":3,"n_ran":2,"n_unverified":1,"n_pointer_only":0}},{"url":"/paper/temporal-interlacing-network","title":"Temporal Interlacing Network","date":"2020-01-17","arxiv_id":"2001.06499","repositories_listed":4,"syntology":{"n":25,"n_ran":16,"n_unverified":9,"n_pointer_only":0}},{"url":"/paper/dvc-an-end-to-end-deep-video-compression","title":"DVC: An End-to-end Deep Video Compression Framework","date":"2018-11-30","arxiv_id":"1812.00101","repositories_listed":4,"syntology":{"n":4,"n_ran":3,"n_unverified":1,"n_pointer_only":3}},{"url":"/paper/locally-consistent-deformable-convolution","title":"Learning Motion in Feature Space: Locally-Consistent Deformable Convolution Networks for Fine-Grained Action Detection","date":"2018-11-21","arxiv_id":"1811.08815","repositories_listed":4,"syntology":null},{"url":"/paper/dgc-net-dense-geometric-correspondence","title":"DGC-Net: Dense Geometric Correspondence Network","date":"2018-10-19","arxiv_id":"1810.08393","repositories_listed":4,"syntology":null},{"url":"/paper/youtube-vos-sequence-to-sequence-video-object","title":"YouTube-VOS: Sequence-to-Sequence Video Object Segmentation","date":"2018-09-03","arxiv_id":"1809.00461","repositories_listed":4,"syntology":null},{"url":"/paper/liteflownet-a-lightweight-convolutional","title":"LiteFlowNet: A Lightweight Convolutional Neural Network for Optical Flow Estimation","date":"2018-05-18","arxiv_id":"1805.07036","repositories_listed":4,"syntology":null},{"url":"/paper/im2flow-motion-hallucination-from-static","title":"Im2Flow: Motion Hallucination from Static Images for Action Recognition","date":"2017-12-12","arxiv_id":"1712.04109","repositories_listed":4,"syntology":null},{"url":"/paper/video-enhancement-with-task-oriented-flow","title":"Video Enhancement with Task-Oriented Flow","date":"2017-11-24","arxiv_id":"1711.09078","repositories_listed":4,"syntology":null},{"url":"/paper/deep-learning-for-precipitation-nowcasting-a","title":"Deep Learning for Precipitation Nowcasting: A Benchmark and A New Model","date":"2017-06-12","arxiv_id":"1706.03458","repositories_listed":4,"syntology":{"n":10,"n_ran":9,"n_unverified":1,"n_pointer_only":5}}],"syntology_records":19,"syntology_note":"a paper without a record is not a recorded non-run: it may lack an arXiv id or simply be absent from the graph layer"},"description_links":{"kept":0,"unwrapped_to_text":0,"bare_urls_linked":0,"relative_images_dropped":0,"rule":"internal links are kept only when the target slug exists in the catalog"},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per-sample execution status on synthesized fixtures ('ran N of M samples'); not a correctness claim and not a ranking signal.","status_vocabulary":{"ran_honours":"ran, honoured the contract we drafted","ran_violates":"ran, violated the contract we drafted","ran_draft_wrong":"ran; our contract draft was wrong, not the code","ran_fixture":"ran; our fixture could not drive it","ran":"ran on a synthesized input","unverified":"unverified (harvested, no recorded run)"}},"not_shown":{"libraries":"the archive has no per-task library table","trend_sparklines":"the Trend column of the benchmarks table was a rendered image; it is not in the archive","social_and_latest_sorts":"stars and social signals are not in the archive"}}