Research map / Video / Video segmentation and tracking
Video Instance Segmentation
Linjie Yang, Yuchen Fan, Ning Xu
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 category consistency cue to the matching score (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 24.7 -> 30.3 | Improved |
| add box IoU cue to the matching score (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 29.2 -> 30.3 | Improved |
| add detection confidence cue to the matching score (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 29.5 -> 30.3 | Improved |
| add all three cues over appearance-only matching (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 21.1 -> 30.3 | Improved |
| add category cue alone to appearance-only matching (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 21.1 -> 27.9 | Improved |
| add box IoU cue alone to appearance-only matching (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 21.1 -> 22.7 | Improved |
| add detection score alone to appearance-only matching (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 21.1 -> 23.4 | Improved |
| replace predicted identities with ground truth identities (Identity Oracle) (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 30.3 -> 31.5 | No clear effect |
| replace predicted image-level detections with ground truth (Image Oracle) (ablation) | MaskTrack R-CNN · YouTube-VIS validation | AP: 30.3 -> 78.7 | Improved |
| replace tracking-by-detection baseline SeqTracker with end-to-end tracking branch | MaskTrack R-CNN · YouTube-VIS test | AP: 29.5 -> 32.3 | Improved |
| replace IoUTracker+ with learned appearance tracking branch | MaskTrack R-CNN · YouTube-VIS validation | AP: 23.6 -> 30.3 | Improved |
| replace mask propagation (FEELVOS) with tracking branch | MaskTrack R-CNN · YouTube-VIS validation | AP: 26.9 -> 30.3 | Improved |
Closest papers by meaning
- End-to-End Video Instance Segmentation With Transformers (CVPR 2021 · similarity 0.92)
- Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation (CVPR 2020 · similarity 0.91)
- Video Mask Transfiner for High-Quality Video Instance Segmentation (ECCV 2022 · similarity 0.90)
- Robust and Consistent Online Video Instance Segmentation via Instance Mask Propagation (AAAI 2025 · similarity 0.89)
- Minimizing Labeled, Maximizing Unlabeled: An Image-Driven Approach for Video Instance Segmentation (CVPR 2025 · similarity 0.89)
- VideoCutLER: Surprisingly Simple Unsupervised Video Instance Segmentation (CVPR 2024 · 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.
