Research map / Video / Video segmentation and tracking
Video Object Segmentation with Joint Re-identification and Attention-Aware Mask Propagation
Xiaoxiao Li, Chen Change Loy
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 attention mechanism to Re-MP (ablation) | Re-MP · DAVIS17 val | G-mean: 67.5 -> 69.1 | Improved |
| add attention mechanism to Re-MP (ablation) | Re-MP · DAVIS17 test-dev | G-mean: 58.0 -> 61.0 | Improved |
| replace MSK with Re-MP (recurrent propagation, RoIAlign) (ablation) | MSK · DAVIS17 val | G-mean: 65.3 -> 69.1 | Improved |
| replace MSK with Re-MP with attention (ablation) | MSK · DAVIS17 test-dev | G-mean: 51.7 -> 61.0 | Improved |
| add Re-ID module with template expansion to Re-MP (ablation) | DyeNet · DAVIS17 test-dev | G-mean: 61.0 -> 68.2 | Improved |
| add online training to offline DyeNet (ablation) | DyeNet · DAVIS17 test-dev | G-mean: 62.5 -> 68.2 | Improved |
| iterate Re-ID with template expansion (iteration 1 to 3+) (ablation) | DyeNet · DAVIS17 val (threshold 0.9) | G-mean: 72.3 -> 73.6 | Improved |
| lower Re-ID threshold from 0.9 to 0.7 (ablation) | DyeNet · DAVIS17 val (iter 3+) | G-mean: 73.6 -> 74.1 | Improved |
| lower Re-ID threshold from 0.7 to 0.6 (ablation) | DyeNet · DAVIS17 val (iter 3+) | G-mean: 74.1 -> 73.9 | Hurt |
| add online fine-tuning to offline DyeNet (ablation) | DyeNet · DAVIS16 | mIoU: 84.7 -> 86.2 | Improved |
| add online fine-tuning to offline DyeNet (ablation) | DyeNet · YouTubeObjects | mIoU: 74.9 -> 79.6 | Improved |
| add online fine-tuning to offline DyeNet (ablation) | DyeNet · SegTrack v2 | mIoU: 78.3 -> 78.7 | No clear effect |
Closest papers by meaning
- Learning Video Object Segmentation With Visual Memory (ICCV 2017 · similarity 0.90)
- Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation (CVPR 2020 · similarity 0.90)
- Target-Aware Object Discovery and Association for Unsupervised Video Multi-Object Segmentation (CVPR 2021 · similarity 0.89)
- MaskRNN: Instance Level Video Object Segmentation (NeurIPS 2017 · similarity 0.89)
- Learning to Segment Moving Objects (IJCV 2018 · similarity 0.89)
- Learning Dynamic Network Using a Reuse Gate Function in Semi-Supervised Video Object Segmentation (CVPR 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.
