Research map / Video / Video segmentation and tracking
VRSTC: Occlusion-Free Video Person Re-Identification
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 non-local blocks to the baseline re-ID network (ablation) | ResNet-50 baseline · iLIDS-VID | rank-1 | Helped |
| add a spatial structure generator trained with spatial reconstruction loss (ablation) | NL re-ID network · iLIDS-VID, MARS, and DukeMTMC-VideoReID | rank-1: +1.3%, +0.9%, and +1.1%, respectively | Helped |
| add the temporal generator to the spatial generator (ablation) | Spa completion model · video re-ID benchmarks | rank-1 accuracy | Helped |
| replace temporal attention generator with a parameter-matched standard autoencoder (ablation) | Spa completion model · video re-ID benchmarks | rank-1 accuracy | No clear effect |
| remove the temporal attention layer from the temporal generator (ablation) | Spa+Tem completion model · video re-ID benchmarks | rank-1 accuracy | Hurt |
| add local and global discriminators to the spatial-temporal generators (ablation) | Spa+Tem completion model · video re-ID benchmarks | rank-1 accuracy | Helped |
| add the ID guider loss to the completion model (ablation) | Spa+Tem with discriminators · video re-ID benchmarks | rank-1 accuracy | Helped |
| set the occlusion threshold to 0.89 rather than 0 (ablation) | similarity scoring mechanism · iLIDS-VID | rank-1: 79.8 -> 80.3 | Helped |
| set the occlusion threshold to 0.89 rather than 0 (ablation) | similarity scoring mechanism · MARS | rank-1: 84.4 -> 84.9 | Helped |
| set the occlusion threshold to 0.89 rather than 0 (ablation) | similarity scoring mechanism · DukeMTMC-VideoReID | rank-1: 91.4 -> 91.7 | Helped |
| increase the occlusion threshold beyond 0.89 (ablation) | similarity scoring mechanism · iLIDS-VID | rank-1: 80.3 -> 78.8 (0.91); 80.3 -> 78.3 (0.93) | Hurt |
| increase the occlusion threshold beyond 0.89 (ablation) | similarity scoring mechanism · MARS | rank-1: 84.9 -> 84.2 (0.91); 84.9 -> 83.6 (0.93) | Hurt |
Closest papers by meaning
- Spatial-Temporal Graph Convolutional Network for Video-Based Person Re-Identification (CVPR 2020 · similarity 0.89)
- Spatial and Temporal Mutual Promotion for Video-Based Person Re-Identification (AAAI 2019 · similarity 0.88)
- Attention Disturbance and Dual-Path Constraint Network for Occluded Person Re-identification (AAAI 2024 · similarity 0.87)
- Dynamic Feature Pruning and Consolidation for Occluded Person Re-identification (AAAI 2024 · similarity 0.86)
- Occluded Person Re-identification via Saliency-Guided Patch Transfer (AAAI 2024 · similarity 0.86)
- Temporal Correlation Vision Transformer for Video Person Re-Identification (AAAI 2024 · similarity 0.85)
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.