Papers › MobileVOS: Real-Time Video Object Segmentation Contrastive Learning meets Knowledge...

MobileVOS: Real-Time Video Object Segmentation Contrastive Learning meets Knowledge Distillation

14 Mar 2023CVPR 2023 1arXiv:2303.07815archive 2025-07-28

Roy Miles, Mehmet Kerim Yucel, Bruno Manganelli, Albert Saa-Garriga

This paper tackles the problem of semi-supervised video object segmentation on resource-constrained devices, such as mobile phones. We formulate this problem as a distillation task, whereby we demonstrate that small space-time-memory networks with finite memory can achieve competitive results with state of the art, but at a fraction of the computational cost (32 milliseconds per frame on a Samsung Galaxy S22). Specifically, we provide a theoretically grounded framework that unifies knowledge distillation with supervised contrastive representation learning. These models are able to jointly benefit from both pixel-wise contrastive learning and distillation from a pre-trained teacher. We validate this loss by achieving competitive J&F to state of the art on both the standard DAVIS and YouTube benchmarks, despite running up to 5x faster, and with 32x fewer parameters.

PaperPDFConference PDF

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

No code repository is listed for this paper in the archive or in Syntology's graph.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Contrastive LearningKnowledge DistillationRepresentation LearningSemantic SegmentationSemi-Supervised Video Object SegmentationVideo Object SegmentationVideo Semantic Segmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS (BL30K) F-measure (Mean) 92.6 #20 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS (BL30K) J&F 91.4 #20 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS (BL30K) Jaccard (Mean) 90.3 #20 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS (BL30K) Speed (FPS) 100.1 #20 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS F-measure (Mean) 91.6 #26 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS J&F 90.6 #26 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS Jaccard (Mean) 89.7 #26 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2016 MobileVOS Speed (FPS) 100.1 #26 of 78 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS (BL30K) F-measure (Mean) 88.9 #38 of 81 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS (BL30K) J&F 82.3 #38 of 81 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS (BL30K) Params(M) 8.1 #38 of 81 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS (BL30K) Speed (FPS) 90.6 #38 of 81 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS F-measure (Mean) 87.1 #45 of 81 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS J&F 80.2 #45 of 81 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS Params(M) 8.1 #45 of 81 Archive leaderboard report
Semi-Supervised Video Object Segmentation DAVIS 2017 (val) MobileVOS Speed (FPS) 90.6 #45 of 81 Archive leaderboard report
Video Object Segmentation DAVIS 2016 MobileVOS (val) F-Score 92.6 #6 of 24 Archive leaderboard report
Video Object Segmentation DAVIS 2016 MobileVOS (val) J&F 91.4 #6 of 24 Archive leaderboard report
Video Object Segmentation DAVIS 2016 MobileVOS (val) Jaccard (Mean) 90.3 #6 of 24 Archive leaderboard report
Video Object Segmentation YouTube-VOS 2019 MobileVOS F-Measure (Seen) 87.7 #6 of 10 Archive leaderboard report
Video Object Segmentation YouTube-VOS 2019 MobileVOS F-Measure (Unseen) 85.3 #6 of 10 Archive leaderboard report
Video Object Segmentation YouTube-VOS 2019 MobileVOS Jaccard (Seen) 83.2 #6 of 10 Archive leaderboard report
Video Object Segmentation YouTube-VOS 2019 MobileVOS Jaccard (Unseen) 76.9 #6 of 10 Archive leaderboard report
Video Object Segmentation YouTube-VOS 2019 MobileVOS Mean Jaccard & F-Measure 83.3 #6 of 10 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Contrastive LearningKnowledge Distillation

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections