Research map / Video / Video segmentation and tracking
InterRVOS: Interaction-Aware Referring Video Object Segmentation
Woojeong Jin, Seongchan Kim, Jaeho Lee, Seungryong Kim
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 interaction-aware special tokens [SEG_ACT] and [SEG_TAR] (ablation) | ReVIOSa-1B · InterRVOS-127K eval | J&F: 57.0 -> 59.6 | Improved |
| add attention mask loss (AML) (ablation) | ReVIOSa-1B · InterRVOS-127K eval | J&F: 57.0 -> 58.5 | Improved |
| combine special tokens with AML (ablation) | ReVIOSa-1B · InterRVOS-127K eval | J&F: 59.6 -> 62.0 | Improved |
| apply AML at layer 22 only (ablation) | ReVIOSa-1B · InterRVOS-127K eval | J&F: +2.4 | Improved |
| apply AML to top-3 vision-attention layers instead of bottom-3 (ablation) | ReVIOSa-1B · InterRVOS-127K eval | J&F | Improved |
| scale ReVIOSa from 1B to 4B | ReVIOSa · InterRVOS-127K eval, Target | J&F: 67.4 -> 68.3 | Improved |
| train on InterRVOS-127K instead of ReVOS (ablation) | baseline MLLM+SAM2 · MeViS valid zero-shot | J&F: 39.6 -> 40.4 | Improved |
| train on InterRVOS-127K instead of Ref-SAV (ablation) | baseline MLLM+SAM2 · MeViS valid zero-shot | J&F: 32.8 -> 40.4 | Improved |
| use ReVIOSa architecture instead of baseline when trained on InterRVOS-127K (ablation) | InterRVOS-127K training · Ref-Youtube-VOS zero-shot | J&F: 61.2 -> 60.3 | Hurt |
| use ReVIOSa architecture instead of baseline when trained on InterRVOS-127K (ablation) | InterRVOS-127K training · MeViS valid zero-shot | J&F: 40.4 -> 42.4 | Improved |
| prompt existing RVOS models to segment the interaction target zero-shot | Sa2VA-4B, VISA-7B, VideoLISA-3.8B · InterRVOS-Target | J&F: 24.2, 19.2, 13.1 versus ReVIOSa-4B 68.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VISA-7B — VISA: Reasoning Video Object Segmentation via Large Language Model (ECCV 2024) · better on 1 of 1 shared result
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
- OnlineRefer: A Simple Online Baseline for Referring Video Object Segmentation (ICCV 2023 · similarity 0.92)
- Temporal Collection and Distribution for Referring Video Object Segmentation (ICCV 2023 · similarity 0.91)
- Long-RVOS: A Comprehensive Benchmark for Long-term Referring Video Object Segmentation (CVPR 2026 · similarity 0.91)
- LoSh: Long-Short Text Joint Prediction Network for Referring Video Object Segmentation (CVPR 2024 · similarity 0.90)
- Semantic and Sequential Alignment for Referring Video Object Segmentation (CVPR 2025 · similarity 0.90)
- SAMWISE: Infusing Wisdom in SAM2 for Text-Driven Video Segmentation (CVPR 2025 · similarity 0.89)
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.
