Research map / Video / Video segmentation and tracking
Co-Segmentation Inspired Attention Networks for Video-Based Person Re-Identification
Arulkumar Subramaniam, Athira Nambiar, Anurag Mittal
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 COSAM after blocks 4 and 5 | ResNet50 + TPavg · MARS | mAP: 75.8 -> 77.2 | Improved |
| add COSAM after blocks 4 and 5 | SE-ResNet50 + TPavg · MARS | mAP: 78.1 -> 79.9 | Improved |
| add COSAM after blocks 4 and 5 | SE-ResNet50 + TA · MARS | mAP: 77.7 -> 79.1 | Improved |
| add COSAM after blocks 4 and 5 | ResNet50 + RNN · DukeMTMC-VideoReID | mAP: 88.1 -> 90.4 | Improved |
| add COSAM after blocks 4 and 5 | SE-ResNet50 + TPavg · iLIDS-VID | rank-1: 76.9 -> 79.6 | Improved |
| add COSAM after blocks 4 and 5 | SE-ResNet50 + RNN · DukeMTMC-VideoReID | rank-1: 94.0 -> 93.9 | No clear effect |
| add COSAM after block 2 only (ablation) | ResNet50 + TPavg · MARS | mAP: 75.8 -> 68.3 | Hurt |
| add COSAM after block 3 only (ablation) | ResNet50 + TPavg · MARS | mAP: 75.8 -> 76.9 | Improved |
| add COSAM after both blocks 4 and 5 instead of block 4 only (ablation) | SE-ResNet50 + TPavg · MARS | mAP: 79.8 -> 79.9 | No clear effect |
| use 4 frames instead of 2 per snippet (ablation) | SE-ResNet50 + COSAM4,5 + TPavg · MARS | mAP: 78.1 -> 79.9 | Improved |
| use 8 frames instead of 4 per snippet (ablation) | SE-ResNet50 + COSAM4,5 + TPavg · DukeMTMC-VideoReID | mAP: 94.1 -> 92.1 | Hurt |
| apply re-ranking | SE-ResNet50 + COSAM4,5 + TPavg · MARS | mAP: 79.9 -> 87.4 | Improved |
Closest papers by meaning
- Leader-Based Multi-Scale Attention Deep Architecture for Person Re-Identification (TPAMI 2019 · similarity 0.89)
- A Multi-Task Deep Network for Person Re-Identification (AAAI 2017 · similarity 0.89)
- Scalable Person Re-Identification by Harmonious Attention (IJCV 2019 · similarity 0.89)
- Deep Learning for Person Re-Identification: A Survey and Outlook (TPAMI 2021 · similarity 0.89)
- READ: Reciprocal Attention Discriminator for Image-to-Video Re-Identification (ECCV 2020 · similarity 0.88)
- Asymmetric Co-Teaching for Unsupervised Cross-Domain Person Re-Identification (AAAI 2020 · 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.
