Research map / Video / Video segmentation and tracking
Super-Trajectory for Video Segmentation
Wenguan Wang, Jianbing Shen, Jianwen Xie, Fatih Porikli
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 |
|---|---|---|---|
| use super-trajectory generation instead of LTM trajectories (ablation) | STV video segmentation · DAVIS | IoU: 0.718 -> 0.736 | Improved |
| use super-trajectory generation instead of DAD trajectories (ablation) | STV video segmentation · DAVIS | IoU: 0.654 -> 0.736 | Improved |
| add reverse tracking of points to treat latecomer trajectories as background (ablation) | STV · DAVIS | IoU: about 6% performance improvement | Improved |
| increase number of spatial grids K from 800 toward 1200 (ablation) | super-trajectory clustering · DAVIS | IoU | Improved |
| increase K beyond 1200 (ablation) | super-trajectory clustering · DAVIS | IoU | No clear effect |
| vary number of nearest neighbors N from 4 to 20 (ablation) | re-occurrence propagation · DAVIS | IoU: +-0.3% performance variation | No clear effect |
| use STV instead of BVS | video segmentation · DAVIS (50 sequences) | IoU: 0.665 -> 0.736 | Improved |
| use STV instead of JOT | video segmentation · SegTrack-V2 | IoU: 0.718 -> 0.781 | Improved |
| use STV instead of HVS on camel | video segmentation · DAVIS camel | IoU: 0.876 -> 0.798 | Hurt |
| use STV instead of BVS on swing | video segmentation · DAVIS swing | IoU: 0.784 -> 0.765 | Hurt |
Closest papers by meaning
- Semi-Supervised Video Object Segmentation with Super-Trajectories (TPAMI 2018 · similarity 0.98)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.91)
- Learned Trajectory Embedding for Subspace Clustering (CVPR 2024 · similarity 0.89)
- Object Discovery in Videos as Foreground Motion Clustering (CVPR 2019 · similarity 0.89)
- Segment Any Motion in Videos (CVPR 2025 · similarity 0.88)
- STEP: Segmenting and Tracking Every Pixel (NeurIPS 2021 · similarity 0.88)
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.
