Research map / Video / Video segmentation and tracking
Multi-Scale 3D Convolution Network for Video Based Person Re-Identification
Read the paperCheck 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 |
|---|---|---|---|
| Replace portions of ResNet50 2D convolutions with M3D layers | ResNet50 on MARS · MARS | mAP: 62.54 -> 69.90 | Helped |
| Use M3D layers instead of P3D-C layers | ResNet50 on MARS, 8-frame input · MARS | mAP: 67.06 -> 69.90 | Helped |
| Add all three RAL branches to M3D (ablation) | M3D on MARS · MARS | mAP: 69.90 -> 71.76 | Helped |
| Add spatial-only RAL to M3D (ablation) | M3D on MARS · MARS | mAP: 69.90 -> 71.04 | Helped |
| Add temporal-only RAL to M3D (ablation) | M3D on MARS · MARS | mAP: 69.90 -> 70.66 | Helped |
| Add channel-only RAL to M3D (ablation) | M3D on MARS · MARS | mAP: 69.90 -> 71.30 | Helped |
| Add all three RAL branches rather than one branch alone (ablation) | M3D on MARS · MARS | mAP: 71.04 -> 71.76; 70.66 -> 71.76; 71.30 -> 71.76 | Helped |
| Fuse 2D spatial features with M3D+RAL temporal features (ablation) | M3D+RAL on MARS · MARS | mAP: 71.76 -> 74.06 | Helped |
| Replace all 2D convolution layers with M3D layers instead of only selected layers (ablation) | M3D CNN on MARS · MARS | mAP | No clear effect |
| Use 16-frame rather than 8-frame clips (ablation) | M3D on MARS · MARS | mAP: 69.90 -> 66.23 | Hurt |
| Use M3D instead of 2D CNN | ResNet50 on PRID · PRID | rank-1 accuracy: 82.02 -> 87.64 | Helped |
| Add all three RAL branches to M3D (ablation) | M3D on PRID · PRID | rank-1 accuracy: 87.64 -> 91.03 | Helped |
Closest papers by meaning
- Temporal-Enhanced Convolutional Network for Person Re-Identification (AAAI 2018 · similarity 0.90)
- A Multi-Task Deep Network for Person Re-Identification (AAAI 2017 · similarity 0.87)
- Efficient and Deep Person Re-Identification Using Multi-Level Similarity (CVPR 2018 · similarity 0.87)
- See the Forest for the Trees: Joint Spatial and Temporal Recurrent Neural Networks for Video-Based Person Re-Identification (CVPR 2017 · similarity 0.87)
- Spatial-Temporal Graph Convolutional Network for Video-Based Person Re-Identification (CVPR 2020 · similarity 0.87)
- Recurrent Convolutional Network for Video-Based Person Re-Identification (CVPR 2016 · similarity 0.87)
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.