Research map / Video / Video segmentation and tracking
Context-Aware Relative Object Queries To Unify Video Instance and Panoptic Segmentation
Anwesa Choudhuri, Girish Chowdhary, Alexander G. Schwing
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 |
|---|---|---|---|
| Propagate and repeatedly refine a single set of object queries across frames (ablation) | Context-aware relative object queries · OVIS validation | AP: 9.8 -> 25.8 | Improved |
| Use propagated queries rather than concatenating learned new-object queries with previous queries (ablation) | Context-aware relative object queries · OVIS validation | AP: 16.9 -> 25.8 | Improved |
| Use relative instead of absolute positional encodings (ablation) | Context-aware relative object queries · KITTI-MOTS car | HOTA: 70.4 -> 83.2 | Improved |
| Use relative instead of absolute positional encodings (ablation) | Context-aware relative object queries · KITTI-MOTS pedestrian | HOTA: 52.0 -> 64.1 | Improved |
| Increase temporal context length from T=1 to T=2 (ablation) | Context-aware relative object queries with ResNet-50 · OVIS validation | AP: 19.1 -> 25.8 | Improved |
| Increase temporal context length from T=1 to T=2 (ablation) | Context-aware relative object queries with Swin-L · OVIS validation | AP: 34.0 -> 38.2 | Improved |
| Increase temporal context length from T=2 to T=4 (ablation) | Context-aware relative object queries with ResNet-50 · OVIS validation | AP: 25.8 -> 25.8 | No clear effect |
| Use the proposed online query propagation method rather than VITA | Video instance segmentation · YouTube-VIS 2019 validation | AP: 49.8 -> 46.7 | Hurt |
| Use the proposed online query propagation method rather than VITA | Video instance segmentation · YouTube-VIS 2021 validation | AP: 45.7 -> 43.3 | Hurt |
| Use the proposed method rather than VPS-Net | Video panoptic segmentation · Cityscapes-VPS | VPQ: 57.5 -> 63.0 | Improved |
Closest papers by meaning
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.89)
- Look Before You Match: Instance Understanding Matters in Video Object Segmentation (CVPR 2023 · similarity 0.89)
- Target-Aware Object Discovery and Association for Unsupervised Video Multi-Object Segmentation (CVPR 2021 · similarity 0.89)
- InsPro: Propagating Instance Query and Proposal for Online Video Instance Segmentation (NeurIPS 2022 · similarity 0.88)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.87)
- A Distractor-Aware Memory for Visual Object Tracking with SAM2 (CVPR 2025 · 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.
