Research map / Detection and segmentation / Open-set and few-shot detection
DyStaB: Unsupervised Object Segmentation via Dynamic-Static Bootstrapping
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 temporal consistency to the mutual-information motion segmentation objective (ablation) | DyStaB dynamic model · DAVIS | mIoU: 59.2 -> 62.4 | Helped |
| Add temporal consistency to the mutual-information motion segmentation objective (ablation) | DyStaB dynamic model · FBMS | mIoU: 36.8 -> 40.0 | Helped |
| Add temporal consistency to the mutual-information motion segmentation objective (ablation) | DyStaB dynamic model · SegTrackV2 | mIoU: 45.6 -> 49.1 | Helped |
| Use confidence-aware adaptation rather than training on all motion masks equally (ablation) | DyStaB static model · DAVIS | mIoU: 73.8 -> 78.2 | Helped |
| Use confidence-aware adaptation rather than training on all motion masks equally (ablation) | DyStaB static model · FBMS | mIoU: 65.5 -> 68.7 | Helped |
| Use confidence-aware adaptation rather than training on all motion masks equally (ablation) | DyStaB static model · SegTrackV2 | mIoU: 65.7 -> 69.3 | Helped |
| Increase bootstrapping from 1 to 2 rounds (ablation) | DyStaB · DAVIS | mIoU: 73.8 -> 79.2 | Helped |
| Increase bootstrapping from 2 to 3 rounds (ablation) | DyStaB · DAVIS | mIoU: 79.2 -> 80.0 | Helped |
| Increase the number of training videos (ablation) | DyStaB static model · Combined saliency test set | mIoU: 48.3 -> 52.4 -> 54.9 -> 58.8 -> 61.4 | Helped |
Closest papers by meaning
- Bootstrapping Objectness From Videos by Relaxed Common Fate and Visual Grouping (CVPR 2023 · similarity 0.92)
- Unsupervised Multi-Object Segmentation by Predicting Probable Motion Patterns (NeurIPS 2022 · similarity 0.90)
- Learning Features by Watching Objects Move (CVPR 2017 · similarity 0.89)
- MOVE: Unsupervised Movable Object Segmentation and Detection (NeurIPS 2022 · similarity 0.89)
- Matching Anything by Segmenting Anything (CVPR 2024 · similarity 0.87)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.86)
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.