{"database":"biostudies-literature","file_versions":[],"scores":null,"additional":{"submitter":["Biderman D"],"funding":["Simons Foundation","NIDDK NIH HHS","Dutch Research Council (NWO)","U.S. Department of Health & Human Services | NIH | National Institute of Neurological Disorders and Stroke (NINDS)","NINDS NIH HHS","National Science Foundation (NSF)","U.S. Department of Health & Human Services | NIH | National Institute of Diabetes and Digestive and Kidney Diseases (National Institute of Diabetes & Digestive & Kidney Diseases)","Wellcome Trust","Gatsby Charitable Foundation"],"pagination":["1316-1328"],"full_dataset_link":["https://www.ebi.ac.uk/biostudies/studies/S-EPMC12087009"],"repository":["biostudies-literature"],"omics_type":["Unknown"],"volume":["21(7)"],"pubmed_abstract":["Contemporary pose estimation methods enable precise measurements of behavior via supervised deep learning with hand-labeled video frames. Although effective in many cases, the supervised approach requires extensive labeling and often produces outputs that are unreliable for downstream analyses. Here, we introduce 'Lightning Pose', an efficient pose estimation package with three algorithmic contributions. First, in addition to training on a few labeled video frames, we use many unlabeled videos and penalize the network whenever its predictions violate motion continuity, multiple-view geometry and posture plausibility (semi-supervised learning). Second, we introduce a network architecture that resolves occlusions by predicting pose on any given frame using surrounding unlabeled frames. Third"],"journal":["Nature methods"],"pubmed_title":["Lightning Pose: improved animal pose estimation via semi-supervised learning, Bayesian ensembling and cloud-native open-source tools."],"pmcid":["PMC12087009"],"funding_grant_id":["216324","5R01DK131086-02","VI.Veni.212.184","NS075023","R01 NS118448","1RF1NS118448-01","K99 NS128075","U19 NS107613","1707398","IOS-2115007","U19 NS123716","K99NS128075","543023","R01 DK131086","GAT3708","R01 NS075023","U19NS123716","RF1 NS118448"],"pubmed_authors":["Bonacchi N","Gercek B","Freitas-Silva L","Mainen ZF","Harris KD","International Brain Laboratory","Boussard J","Laranjeira I","Shi Y","Varol E","Huber F","Noone D","Bayer HM","McMannon B","Chapuis GA","Deneve S","Dayan P","Rau F","Saniee K","Schaeffer R","Rossant C","Meshulam L","Hu F","Pouget A","Vishnubhotla A","Ahmadi M","Martinez-Vergara H","Xu Z","Roth N","Miller K","Socha KZ","Engel T","Haetzel LM","Nylund K","Carandini M","Latham P","Noel JP","Faulkner M","Yu H","Bougrova K","Langdon C","Wool LE","Arlandis J","Amjad J","Churchland AK","Soitu C","Bhagat J","Krasniak CS","West SJ","Mohammadi Z","Biderman D","Ghani N","Meijer GT","Melin M","Dan Y","Krishnagopal S","Warren R","Gordeliy I","Bruijns SA","Oloomi F","Benson B","Benson J","Mrsic-Flogel T","Wells MJ","Aguillon-Rodriguez V","Paninski L","Miska NJ","Windolf C","Lee RS","Hurwitz C","Hiratani N","Acerbi L","Whiteway MR","Makri-Cottington A","Sawtell NB","Catarino JA","Mitelut C","Buchanan EK","Krumin M","Langfield C","Steinmetz NA","Schartner MM","Findling C","Soares C","Pillow JW","Ganguli S","Dong LL","McRoberts I","Khanal A","Taheri M","Birman D","Pedraja F","DeWitt EEJ","Hofer S","Barrell H","Kundu D","Pan-Vazquez A","Picard S","Svoboda K","Landemard A","Murakami M","Roy NA","Fiete I","Tessereau C","Abbot L","Fabbri M","Lee HD","Greenspan N","Cunningham JP","Huntenburg JM","Davatolhagh F","Banga K","Campbell R","Lau P","Winter O","Ashwood ZC","Angelaki D","Cazettes F","Pezzotta A","Zhang Y","Zador AM","Liu A","Fetcho R","Witten I","Hausser M","Urai AE"],"additional_accession":[]},"is_claimable":false,"name":"Lightning Pose: improved animal pose estimation via semi-supervised learning, Bayesian ensembling and cloud-native open-source tools.","description":"Contemporary pose estimation methods enable precise measurements of behavior via supervised deep learning with hand-labeled video frames. Although effective in many cases, the supervised approach requires extensive labeling and often produces outputs that are unreliable for downstream analyses. Here, we introduce 'Lightning Pose', an efficient pose estimation package with three algorithmic contributions. First, in addition to training on a few labeled video frames, we use many unlabeled videos and penalize the network whenever its predictions violate motion continuity, multiple-view geometry and posture plausibility (semi-supervised learning). Second, we introduce a network architecture that resolves occlusions by predicting pose on any given frame using surrounding unlabeled frames. Third","dates":{"release":"2024-01-01T00:00:00Z","publication":"2024 Jul","modification":"2026-05-29T17:11:17.661Z","creation":"2026-04-08T05:26:36.876Z"},"accession":"S-EPMC12087009","cross_references":{"pubmed":["38918605"],"doi":["10.1038/s41592-024-02319-1"]}}