Research map / Video / Video segmentation and tracking
YouTube-VOS: Sequence-to-Sequence Video Object Segmentation
Ning Xu, Linjie Yang, Yuchen Fan, Jianchao Yang, Dingcheng Yue, Yuchen Liang, Brian Price, Scott Cohen, Thomas Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add online learning (ablation) | Ours · YouTube-VOS test (seen) | J mean: 60.9 -> 66.9 | Improved |
| add online learning (ablation) | Ours · DAVIS 2016 val | mean IoU: 76.5 -> 79.1 | Improved |
| increase training data from 25% to 100% (ablation) | Ours (w/o OL) · YouTube-VOS test (seen) | J mean: 46.7 -> 60.9 | Improved |
| increase training data from 25% to 100% (ablation) | Ours (w/o OL) · YouTube-VOS test (unseen) | J mean: 40.1 -> 60.1 | Improved |
| add Initializer network (RGB frame plus mask) instead of using the mask as initial hidden state (ablation) | Ours · YouTube-VOS test (seen) | J mean: 45.1 -> 60.9 | Improved |
| add Initializer network instead of mask as hidden state (ablation) | Ours · YouTube-VOS test (unseen) | J mean: 38.6 -> 60.1 | Improved |
| feed previous predicted mask into Encoder (teacher forcing + curriculum) (ablation) | Ours · YouTube-VOS test (seen) | J mean: 60.9 -> 59.4 | No clear effect |
| feed previous predicted mask into Encoder (teacher forcing + curriculum) (ablation) | Ours · YouTube-VOS test (unseen) | J mean: 60.1 -> 60.7 | No clear effect |
| use pretrained DeepLab as Encoder instead of training on YouTube-VOS (ablation) | Ours · DAVIS 2016 val | mean IoU: 51.3 -> 45.6 | Hurt |
| add SegTrackv2, JumpCut and YoutubeObjects videos to 30 DAVIS training videos (ablation) | Ours (from scratch) · DAVIS 2016 val | mean IoU: 51.3 -> 51.9 | No clear effect |
| train on YouTube-VOS instead of DAVIS-scale data | Ours · DAVIS 2016 val | mean IoU: 51.9 -> 76.5 | Improved |
Closest papers by meaning
- Point-VOS: Pointing Up Video Object Segmentation (CVPR 2024 · similarity 0.91)
- Learning What to Learn for Video Object Segmentation (ECCV 2020 · similarity 0.90)
- LVOS: A Benchmark for Long-term Video Object Segmentation (ICCV 2023 · similarity 0.90)
- YouMVOS: An Actor-Centric Multi-Shot Video Object Segmentation Dataset (CVPR 2022 · similarity 0.90)
- Learning Fast and Robust Target Models for Video Object Segmentation (CVPR 2020 · similarity 0.90)
- LVOS: A Benchmark for Large-Scale Long-Term Video Object Segmentation (TPAMI 2025 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
