Research map / Video / Video segmentation and tracking
Temporal Collection and Distribution for Referring Video Object Segmentation
Jiajin Tang, Ge Zheng, Sibei Yang
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 cross-frame temporal attention on a referent sequence selected from local queries (ablation) | local-query baseline · Ref-Youtube-VOS | J&F: 55.0 -> 57.6 | Improved |
| add global queries that collect motion information and distribute context to local queries (ablation) | local-query baseline · Ref-Youtube-VOS | J&F: 55.0 -> 59.8 | Improved |
| combine cross-frame temporal modeling with motion collection and distribution (ablation) | TempCD · Ref-Youtube-VOS | J&F: 59.8 -> 62.6 | Improved |
| replace Top-K selection with Gumbel Softmax (ablation) | TempCD · Ref-Youtube-VOS | J&F: 62.6 -> 63.4 | Improved |
| let local object queries also receive temporal context from the referent sequence (ablation) | TempCD · Ref-Youtube-VOS | J&F: 63.4 -> 65.8 | Improved |
| use a single shared set of global video-level queries instead of per-frame local queries (ablation) | query baseline · Ref-Youtube-VOS | J&F: 55.0 -> 54.2 | Hurt |
| TempCD versus ReferFormer with ResNet-50 | RVOS decoder · Ref-Youtube-VOS | J&F: 55.6 -> 59.0 | Improved |
| TempCD versus ReferFormer with Video-Swin-B | RVOS decoder · Ref-Davis17 | J&F: 61.1 -> 64.6 | Improved |
| TempCD versus LBDT on A2D-Sentences | RVOS model · A2D-Sentences | oIoU: 70.4 -> 76.6 | Improved |
Closest papers by meaning
- Language As Queries for Referring Video Object Segmentation (CVPR 2022 · similarity 0.92)
- SOC: Semantic-Assisted Object Cluster for Referring Video Object Segmentation (NeurIPS 2023 · similarity 0.92)
- ReferEverything: Towards Segmenting Everything We Can Speak of in Videos (ICCV 2025 · similarity 0.91)
- LoSh: Long-Short Text Joint Prediction Network for Referring Video Object Segmentation (CVPR 2024 · similarity 0.89)
- Semantic and Sequential Alignment for Referring Video Object Segmentation (CVPR 2025 · similarity 0.89)
- 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.
