Research map / Video / Video segmentation and tracking
Temporal Complementary Learning for Video Person Re-Identification
Ruibing Hou, Hong Chang, Bingpeng Ma, Shiguang Shan, Xilin Chen
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 TSE (ablation) | ResNet-50 baseline · MARS | mAP / top-1: 79.6 -> 82.5 mAP; 86.8 -> 88.2 top-1 | Improved |
| add saliency erasing operation to ordered learners (ablation) | TSE · MARS | mAP: 80.9 -> 82.5 | Improved |
| add ordered learners without saliency erasing (ablation) | ResNet-50 baseline · MARS | mAP: 79.6 -> 80.9 | No clear effect |
| add TSB at stage2 (ablation) | ResNet-50 baseline · MARS | mAP / top-1: 79.6 -> 82.3 mAP; 86.8 -> 87.6 top-1 | Improved |
| add TSB to TSE (ablation) | base.+TSE · MARS | mAP / top-1: 82.5 -> 83.0 mAP; 88.2 -> 88.8 top-1 | Improved |
| add second TSB module at stage3 in addition to stage2 (ablation) | TCLNet · MARS | mAP: 83.0 -> 82.7 | No clear effect |
| increase number of learners N from 2 to 4 (ablation) | base.+TSE · MARS | mAP: 82.5 -> 81.0 | Hurt |
| use erased block height 3 instead of 2, 4 or 5 (ablation) | base.+TSE · MARS | mAP: 81.8, 82.0, 81.7 -> 82.5 | Improved |
| add TSE instead of DropBlock (ablation) | ResNet-50 baseline · MARS | mAP / top-1: 79.8 -> 82.5 mAP; 86.9 -> 88.2 top-1 | Improved |
| add Random Erasing on top of TSE (ablation) | base.+TSE · MARS | mAP: 82.5 -> 82.9 | Improved |
| use TSB instead of 3D convolution (ablation) | ResNet-50 baseline · MARS | mAP: 80.0 -> 82.3 | Improved |
| use TSB instead of non-local with TSE (ablation) | base.+TSE · MARS | top-1: 87.6 -> 88.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VRSTC — VRSTC: Occlusion-Free Video Person Re-Identification (CVPR 2019) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Salient-to-Broad Transition for Video Person Re-Identification (CVPR 2022 · similarity 0.87)
- Temporal Complementarity-Guided Reinforcement Learning for Image-to-Video Person Re-Identification (CVPR 2022 · similarity 0.86)
- Rethinking Temporal Fusion for Video-Based Person Re-Identification on Semantic and Time Aspect (AAAI 2020 · similarity 0.86)
- Watching You: Global-Guided Reciprocal Learning for Video-Based Person Re-Identification (CVPR 2021 · similarity 0.86)
- Temporal Correlation Vision Transformer for Video Person Re-Identification (AAAI 2024 · similarity 0.85)
- TF-CLIP: Learning Text-Free CLIP for Video-Based Person Re-identification (AAAI 2024 · similarity 0.84)
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.
