{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,29]],"date-time":"2026-10-29T14:38:14Z","timestamp":1793284694577,"version":"4.5.0"},"reference-count":46,"publisher":"Public Library of Science (PLoS)","issue":"7","license":[{"start":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T00:00:00Z","timestamp":1783382400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Shaanxi University of Technology","award":["SLGRC202416"],"award-info":[{"award-number":["SLGRC202416"]}],"id":[{"id":"https:\/\/ror.org\/056m91h77","id-type":"ROR","asserted-by":"crossref"}]}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>The Segment Anything Model 2 (SAM 2) has emerged as a robust foundational model for video object segmentation. Nevertheless, it exhibits notable limitations in crowded scenarios, particularly those involving fast-moving objects or self-occlusion. Furthermore, its greedy-selection memory architecture is afflicted by \u201cerror accumulation,\u201d which collectively impairs its performance in consistent object segmentation tasks. To address these critical drawbacks and enhance the consistency and robustness of SAM 2 in complex video scenarios, an enhanced variant of SAM 2, termed UAMP, is proposed, which integrates memory propagation based on explicit appearance and motion uncertainty modeling, coupled with a dual mechanism encompassing long-term memory updating and short-term memory selection. By fusing uncertainty-aware representations with these dual memory mechanisms, UAMP effectively accommodates dynamic variations in object appearance and motion, further refines the object memory bank, and thereby realizes consistent video object segmentation. Quantitative and qualitative evaluations conducted on diverse benchmark datasets demonstrate that UAMP achieves superior performance, particularly in scenarios involving occlusions and object reappearances. Specifically, UAMP yields consistent performance improvements over state-of-the-art (SOTA) methods across five video object segmentation (VOS) benchmarks, with a maximum enhancement of 5.6 points in the J&amp;F metric. These findings underscore the robustness and effectiveness of the proposed UAMP method in addressing complex tracking scenarios, thus providing a valuable enhancement to SAM 2 for practical video object segmentation applications.<\/jats:p>","DOI":"10.1371\/journal.pone.0353156","type":"journal-article","created":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T17:36:04Z","timestamp":1783445764000},"page":"e0353156","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["UAMP: Consistent video object segmentation with uncertainty-aware memory propagation"],"prefix":"10.1371","volume":"21","author":[{"given":"Yichuang","family":"Luo","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Fang","family":"Wang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0896-1810","authenticated-orcid":true,"given":"Xiaohu","family":"Liu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,7,7]]},"reference":[{"key":"pone.0353156.ref001","first-page":"28085","article-title":"SAM 2: Segment anything in images and videos.","volume-title":"International Conference on learning representations (ICLR)","author":"N Ravi","year":"2025"},{"key":"pone.0353156.ref002","article-title":"The 2017 davis challenge on video object segmentation","author":"J Pont-Tuset","year":"2017"},{"key":"pone.0353156.ref003","doi-asserted-by":"crossref","first-page":"13614","DOI":"10.1109\/ICCV51701.2025.01264","article-title":"SAM2LONG: enhancing SAM 2 for long video segmentation with a training-free memory tree.","author":"S Ding","year":"2025","journal-title":"2025 IEEE\/CVF International Conference on Computer Vision (ICCV)"},{"key":"pone.0353156.ref004","article-title":"SAMURAI: adapting segment anything model for zero-shot visual tracking with motion-aware memory","author":"CY Yang","year":"2024"},{"key":"pone.0353156.ref005","doi-asserted-by":"crossref","unstructured":"Videnovic J, Lukezic A, Kristan M. A distractor-aware memory for visual object tracking with SAM2. In: 2025 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2025. p. 24255\u201364. https:\/\/doi.org\/10.1109\/cvpr52734.2025.02259","DOI":"10.1109\/CVPR52734.2025.02259"},{"issue":"4","key":"pone.0353156.ref006","doi-asserted-by":"crossref","first-page":"4642","DOI":"10.1109\/TPAMI.2025.3646919","article-title":"DC-SAM: in-context segment anything in images and videos via dual consistency","volume":"48","author":"M Qi","year":"2026","journal-title":"IEEE Trans Pattern Anal Mach Intell"},{"key":"pone.0353156.ref007","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR52734.2025.00684","article-title":"MatAnyone: stable video matting with consistent memory propagation. arXiv","author":"P Yang","year":"2025"},{"key":"pone.0353156.ref008","doi-asserted-by":"crossref","unstructured":"Robinson A, Jaremo Lawin F, Danelljan M, Khan FS, Felsberg M. Learning fast and robust target models for video object segmentation. In: 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2020. p. 7404\u201313. https:\/\/doi.org\/10.1109\/cvpr42600.2020.00743","DOI":"10.1109\/CVPR42600.2020.00743"},{"key":"pone.0353156.ref009","first-page":"794","article-title":"Learning what to learn for video object segmentation","volume":"777","author":"G Bhat","year":"2020","journal-title":"Eur Conf Comput Vis"},{"key":"pone.0353156.ref010","article-title":"Online adaptation of convolutional neural networks for video object segmentation. arXiv","author":"P Voigtlaender","year":"2017"},{"key":"pone.0353156.ref011","doi-asserted-by":"crossref","unstructured":"Chen Y, Pont-Tuset J, Montes A. Blazingly fast video object segmentation with pixel-wise metric learning. In: Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), USA. 2018. p. 1189\u201398.","DOI":"10.1109\/CVPR.2018.00130"},{"key":"pone.0353156.ref012","first-page":"348","article-title":"Collaborative Video Object Segmentation by Foreground-Background Integration. In Computer Vision \u2013 ECCV 2020","volume":"332","author":"Z Yang","journal-title":"Lecture Notes in Computer Science"},{"key":"pone.0353156.ref013","doi-asserted-by":"crossref","first-page":"619","DOI":"10.1007\/978-3-030-01228-1_36","article-title":"YouTube-VOS: sequence-to-sequence video object segmentation","volume-title":"Computer Vision \u2013 ECCV 2018","author":"N Xu","year":"2018"},{"key":"pone.0353156.ref014","doi-asserted-by":"crossref","unstructured":"Wu Q, Yang T, Wu W. Scalable video object segmentation with simplified framework. In: 2023 IEEE\/CVF International Conference on Computer Vision (ICCV). Paris, France, 2023. p. 13833\u201343.","DOI":"10.1109\/ICCV51070.2023.01276"},{"key":"pone.0353156.ref015","doi-asserted-by":"crossref","unstructured":"Goyal R, Fan W-C, Siam M, Sigal L. TAM-VT: transformation-aware multi-scale video transformer for segmentation and tracking. In: 2025 IEEE\/CVF Winter Conference on Applications of Computer Vision (WACV). 2025. p. 8336\u201345. https:\/\/doi.org\/10.1109\/wacv61041.2025.00808","DOI":"10.1109\/WACV61041.2025.00808"},{"key":"pone.0353156.ref016","doi-asserted-by":"crossref","unstructured":"Tokmakov P, Li J, Gaidon A. Breaking the \u201cObject\u201d in Video Object Segmentation. In: 2023 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2023. p.22836\u201345. https:\/\/doi.org\/10.1109\/cvpr52729.2023.02187","DOI":"10.1109\/CVPR52729.2023.02187"},{"key":"pone.0353156.ref017","doi-asserted-by":"crossref","first-page":"282","DOI":"10.1007\/978-3-642-15555-0_21","article-title":"Object segmentation by long term analysis of point trajectories","author":"T Brox","year":"2010","journal-title":"Lecture Notes in Computer Science. Springer Berlin Heidelberg"},{"key":"pone.0353156.ref018","article-title":"LVOS: a benchmark for large-scale long-term video object segmentation","author":"L Hong","year":"2024"},{"key":"pone.0353156.ref019","doi-asserted-by":"crossref","unstructured":"Ding H, Liu C, He S. MOSE: a new dataset for video object segmentation in complex scenes. In: 2023 IEEE\/CVF International Conference on Computer Vision (ICCV), Paris, France, 2023. p. 20167\u201377.","DOI":"10.1109\/ICCV51070.2023.01850"},{"key":"pone.0353156.ref020","doi-asserted-by":"crossref","unstructured":"Cheng H, Schwing AG. XMem: long-term video object segmentation with an Atkinson-Shiffrin memory model. In European Conference on Computer Vision, Tel Aviv, Israel, 23-27 Oct. 2022; 640:658.","DOI":"10.1007\/978-3-031-19815-1_37"},{"key":"pone.0353156.ref021","doi-asserted-by":"crossref","unstructured":"Zhou J, Pang Z, Wang YX. RMem: Restricted Memory banks improve video object segmentation. In: 2024 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA. 2024. p. 18602\u201311.","DOI":"10.1109\/CVPR52733.2024.01760"},{"key":"pone.0353156.ref022","doi-asserted-by":"crossref","unstructured":"Cheng HK, Oh SW, Price B, Lee J-Y, Schwing A. Putting the object back into video object segmentation. In: 2024 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2024. p. 3151\u201361. https:\/\/doi.org\/10.1109\/cvpr52733.2024.00304","DOI":"10.1109\/CVPR52733.2024.00304"},{"key":"pone.0353156.ref023","doi-asserted-by":"crossref","unstructured":"Zhou S, Zhang J, Pan J, Zuo W, Xie H, Ren J. Spatio-temporal filter adaptive network for video deblurring. In: 2019 IEEE\/CVF International Conference on Computer Vision (ICCV). 2019. p. 2482\u201391. https:\/\/doi.org\/10.1109\/iccv.2019.00257","DOI":"10.1109\/ICCV.2019.00257"},{"key":"pone.0353156.ref024","doi-asserted-by":"crossref","unstructured":"Zhou S, Yang P, Wang J. Upscale-a-video: temporal-consistent diffusion model for real-world video super-resolution. In: 2024 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA. 2024. p. 2535\u201345.","DOI":"10.1109\/CVPR52733.2024.00245"},{"key":"pone.0353156.ref025","unstructured":"Xiao C, Cao Q, Zhong Y. MotionTrack: learning motion predictor for multiple object tracking. In: 2023 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, Canada. 2023. p. 17939\u201348."},{"key":"pone.0353156.ref026","article-title":"Window attention is bugged: how not to interpolate position embeddings","author":"D Bolya","year":"2023"},{"key":"pone.0353156.ref027","unstructured":"Katharopoulos A, Vyas A, Pappas N. Transformers are RNNs: fast autoregressive transformers with linear attention. In: Proceedings of the 37th International Conference on Machine Learning. 2020. p. 5156\u201365."},{"key":"pone.0353156.ref028","article-title":"Retentive network: a successor to transformer for large language models","author":"Y Sun","year":"2023"},{"key":"pone.0353156.ref029","article-title":"YouTube-VOS: a large-scale video object segmentation benchmark. arXiv","author":"N Xu","year":"2018"},{"key":"pone.0353156.ref030","first-page":"2491","article-title":"Associating objects with transformers for video object segmentation","volume":"34","author":"Z Yang","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"pone.0353156.ref031","article-title":"Rotary position embedding for vision transformer","author":"B Song","year":"2024"},{"issue":"34","key":"pone.0353156.ref032","first-page":"11781","article-title":"Rethinking space-time networks with improved memory coverage for efficient video object segmentation","volume":"34","author":"H Cheng","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"pone.0353156.ref033","doi-asserted-by":"crossref","unstructured":"Yang Z, Yang Y. Decoupling features in hierarchical propagation for video object segmentation. In: Advances in Neural Information Processing Systems 35, 2022. p. 36324\u201336. https:\/\/doi.org\/10.52202\/068431-2632","DOI":"10.52202\/068431-2632"},{"key":"pone.0353156.ref034","doi-asserted-by":"crossref","unstructured":"Li M, Hu L, Xiong Z. Recurrent dynamic embedding for video object segmentation. In: 2022 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, USA. 2022. p. 1322\u201331.","DOI":"10.1109\/CVPR52688.2022.00139"},{"key":"pone.0353156.ref035","doi-asserted-by":"crossref","unstructured":"Cheng HK, Wug Oh S, Price B, Schwing A, Lee J-Y. Tracking anything with decoupled video segmentation. In: 2023 IEEE\/CVF International Conference on Computer Vision (ICCV), 2023. p. 1316\u201326. https:\/\/doi.org\/10.1109\/iccv51070.2023.00127","DOI":"10.1109\/ICCV51070.2023.00127"},{"key":"pone.0353156.ref036","doi-asserted-by":"crossref","unstructured":"Liu Q, Wang J, Yang Z, Li L. LiVOS: light video object segmentation with gated linear matching. In: Proceedings of the IEEE\/CVF International Conference on Computer Vision and Pattern Recognition (CVPR). 2025. p. 8668\u201378.","DOI":"10.1109\/CVPR52734.2025.00810"},{"key":"pone.0353156.ref037","doi-asserted-by":"crossref","unstructured":"Hong L, Chen W, Liu Z. LVOS: a benchmark for long-term video object segmentation. In: 2023 IEEE\/CVF International Conference on Computer Vision (ICCV), Paris, France. 2023. p. 13434\u201346.","DOI":"10.1109\/ICCV51070.2023.01240"},{"key":"pone.0353156.ref038","doi-asserted-by":"crossref","unstructured":"Fan H, Lin L, Yang F, Chu P, Deng G. LaSOT: a high-quality benchmark for large-scale single object tracking. In: 2019 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA. 2019. p. 5369\u201378.","DOI":"10.1109\/CVPR.2019.00552"},{"issue":"1","key":"pone.0353156.ref039","doi-asserted-by":"crossref","first-page":"439","DOI":"10.1007\/s11263-020-01387-y","article-title":"LaSOT: a high-quality large-scale single object tracking benchmark","volume":"129","author":"H Fan","year":"2021","journal-title":"International Journal of Computer Vision"},{"issue":"5","key":"pone.0353156.ref040","doi-asserted-by":"crossref","first-page":"1562","DOI":"10.1109\/TPAMI.2019.2957464","article-title":"GOT-10k: a large high-diversity benchmark for generic object tracking in the wild","volume":"43","author":"L Huang","year":"2021","journal-title":"IEEE Trans Pattern Anal Mach Intell"},{"key":"pone.0353156.ref041","doi-asserted-by":"crossref","unstructured":"Mayer C, Danelljan M, Pani Paudel D, Van Gool L. Learning target candidate association to keep track of what not to track. In: 2021 IEEE\/CVF International Conference on Computer Vision (ICCV), Montreal, QC, Canada. 2021. p. 13444\u201354.","DOI":"10.1109\/ICCV48922.2021.01319"},{"key":"pone.0353156.ref042","doi-asserted-by":"crossref","unstructured":"Wu Q, Yang T, Liu Z, Wu B, Shan Y, Chan AB. DropMAE: masked autoencoders with spatial-attention dropout for tracking tasks. In: 2023 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2023. p. 14561\u201371. https:\/\/doi.org\/10.1109\/cvpr52729.2023.01399","DOI":"10.1109\/CVPR52729.2023.01399"},{"key":"pone.0353156.ref043","doi-asserted-by":"crossref","unstructured":"Chen X, Peng H, Wang D, Lu H, Hu H. SeqTrack: sequence to sequence learning for visual object tracking. In: 2023 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2023. p. 14572\u201381. https:\/\/doi.org\/10.1109\/cvpr52729.2023.01400","DOI":"10.1109\/CVPR52729.2023.01400"},{"key":"pone.0353156.ref044","doi-asserted-by":"crossref","unstructured":"Cai Y, Liu J, Tang J, Wu G. Robust object modeling for visual tracking. In: 2023 IEEE\/CVF International Conference on Computer Vision (ICCV). 2023. p. 9555\u201366. https:\/\/doi.org\/10.1109\/iccv51070.2023.00879","DOI":"10.1109\/ICCV51070.2023.00879"},{"issue":"7","key":"pone.0353156.ref045","doi-asserted-by":"crossref","first-page":"7588","DOI":"10.1609\/aaai.v38i7.28591","article-title":"ODTrack: online dense temporal token learning for visual tracking","volume":"38","author":"Y Zheng","year":"2024","journal-title":"AAAI"},{"key":"pone.0353156.ref046","doi-asserted-by":"crossref","unstructured":"Lin L, Fan H, Zhang Z, Wang Y, Xu Y. Tracking meets LoRA: faster training, larger model, stronger performance. In: European Conference on Computer Vision, Honolulu, Hawai\u2019I. 2025. p. 318.","DOI":"10.1007\/978-3-031-73232-4_17"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0353156","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,29]],"date-time":"2026-10-29T14:30:19Z","timestamp":1793284219000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0353156"}},"subtitle":[],"editor":[{"given":"Taikyeong Ted","family":"Jeong","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,7,7]]},"references-count":46,"journal-issue":{"issue":"7","published-online":{"date-parts":[[2026,7,7]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0353156","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,7]]}}}