Research map / Video / Video segmentation and tracking
Segment Every Reference Object in Spatial and Temporal Spaces
Jiannan Wu, Yi Jiang, Bin Yan, Huchuan Lu, Zehuan Yuan, Ping Luo
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 |
|---|---|---|---|
| multi-task joint training instead of single-task training (ablation) | UniRef++-R50 · Ref-Youtube-VOS | J&F: 58.3 -> 60.1 | Improved |
| multi-task joint training instead of single-task training (ablation) | UniRef++-R50 · RefCOCO | oIoU: 72.1 -> 74.7 | Improved |
| multi-task joint training instead of single-task training (ablation) | UniRef++-R50 · FSS-1000 | mIoU: 81.8 -> 80.3 | Hurt |
| share one UniFusion module across tasks (vs per-level-only) (ablation) | UniFusion · FSS-1000 | mIoU: 75.2 -> 80.3 | Improved |
| share one UniFusion module across tasks and levels (vs task-shared only) (ablation) | UniFusion · FSS-1000 | mIoU: 64.1 -> 80.3 | Improved |
| share UniFusion across visual levels (ablation) | UniFusion · Youtube-VOS2018 | G: 80.9 -> 81.9 | Improved |
| increase object queries from 1 to 100 (ablation) | UniRef++-R50 · Youtube-VOS2018 | G: 79.0 -> 81.9 | Improved |
| increase object queries from 100 to 300 (ablation) | UniRef++-R50 · RefCOCO | oIoU: 75.1 -> 74.7 | No clear effect |
| add mask reference (previous-frame mask propagation) for RVOS (ablation) | UniRef++-R50 · Ref-Youtube-VOS | J&F: 60.4 -> 61.5 | Improved |
| add mask reference for RVOS (ablation) | UniRef++-L · Ref-Youtube-VOS | J&F: 65.5 -> 66.9 | Improved |
| use both first and previous frame as references (ablation) | UniRef++-R50 · Youtube-VOS2018 | G: 80.0 -> 81.9 | Improved |
| add mask reference for RVOS on Ref-DAVIS17 (UniRef++ vs language-only UniRef) | UniRef++-L · Ref-DAVIS17 | J&F: UniRef-R50 63.5 vs UniRef++-R50 62.5 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- XMem — XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model (ECCV 2022) · better on 0 of 2 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
- Few-Shot Referring Video Single- and Multi-Object Segmentation Via Cross-Modal Affinity with Instance Sequence Matching (IJCV 2025 · similarity 0.90)
- Hybrid Global-Local Representation with Augmented Spatial Guidance for Zero-Shot Referring Image Segmentation (CVPR 2025 · similarity 0.88)
- UniVS: Unified and Universal Video Segmentation with Prompts as Queries (CVPR 2024 · similarity 0.88)
- Mask Grounding for Referring Image Segmentation (CVPR 2024 · similarity 0.87)
- Referred by Multi-Modality: A Unified Temporal Transformer for Video Object Segmentation (AAAI 2024 · similarity 0.87)
- Segment Anything without Supervision (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.
