Research map / Video / Video-language understanding
VideoSeg-R1:Reasoning Video Object Segmentation via Reinforcement Learning
Zishan Xu, Yifu Guo, Yuquan Lu, Fengyu Yang, Junxin Li, Lihua Cai
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 |
|---|---|---|---|
| train with RL instead of SFT (ablation) | VideoSeg-R1 (Qwen2.5-VL-7B) · ReVOS | J&F: 50.9 -> 58.9 | Improved |
| train with RL instead of SFT (ablation) | VideoSeg-R1 · ReasonSeg | gIoU: 59.4 -> 67.2 | Improved |
| add CoT reasoning reward to RL (ablation) | VideoSeg-R1 · ReVOS | J&F: 58.9 -> 61.1 | Improved |
| add CoT reasoning reward to RL (ablation) | VideoSeg-R1 · ReasonSeg | gIoU: 67.2 -> 69.8 | Improved |
| use hierarchical key-segment + target-frame sampler instead of first frame (ablation) | VideoSeg-R1 · MeViS | J&F: 45.7 -> 55.3 | Improved |
| add key-segment stage on top of target-frame localization (ablation) | VideoSeg-R1 · ReVOS | J&F: 57.6 -> 61.1 | Improved |
| use random frame from key segment instead of target-frame localization (ablation) | VideoSeg-R1 · MeViS | J&F: 55.3 -> 49.2 | Hurt |
| add bounding box prompt over no-prompt baseline (ablation) | VideoSeg-R1 · ReVOS | J&F: 40.2 -> 57.2 | Improved |
| add central point to Bbox (ablation) | VideoSeg-R1 · ReasonSeg | gIoU: 65.7 -> 68.2 | Improved |
| add negative point to Bbox + Pcentral (ablation) | VideoSeg-R1 · ReVOS | J&F: 60.3 -> 61.1 | Improved |
| add soft length penalty (ablation) | VideoSeg-R1 · Ref-DAVIS17 | J&F: 75.2 -> 79.8 | Improved |
| add soft length penalty (ablation) | VideoSeg-R1 · ReVOS | reasoning tokens: 89.9 -> 56.1 | Improved |
Later papers that compared with this method
1 later accepted paper reports VideoSeg-R1 as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Beyond Token Representations: Explicit Visual Object Grounding for Video Reasoning Segmentation (NeurIPS 2026 · better on 3 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Video-R1: Reinforcing Video Reasoning in MLLMs (NeurIPS 2025 · similarity 0.91)
- Fast Reasoning Segmentation for Images and Videos (CVPR 2026 · similarity 0.89)
- The Devil is in Temporal Token: High Quality Video Reasoning Segmentation (CVPR 2025 · similarity 0.87)
- SAM-R1: Leveraging SAM for Reward Feedback in Multimodal Segmentation via Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- LENS: Learning to Segment Anything with Unified Reinforced Reasoning (AAAI 2026 · similarity 0.87)
- One Token to Seg Them All: Language Instructed Reasoning Segmentation in Videos (NeurIPS 2024 · similarity 0.87)
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.
