Research map / Label-efficient and robust learning / Recognition and long-tail
MSINet: Twins Contrastive Search of Multi-Scale Interaction for Object ReID
Jianyang Gu, Kai Wang, Hao Luo, Chen Chen, Wei Jiang, Yuqiang Fang, Shanghang Zhang, Yang You, Jian Zhao
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 |
|---|---|---|---|
| Use a partly overlapping identity split (60% train / 80% valid) rather than fully overlapping identities (ablation) | TCM architecture search · MSMT17 | mAP: 50.1 -> 52.5 | Improved |
| Replace softmax classification search with TCM contrastive supervision and identity separation (ablation) | MSINet search · MSMT17 | mAP: 49.8 -> 52.5 | Improved |
| Add Spatial Alignment Module using PAM-Self for positive and negative alignment (ablation) | MSINet · VeRi-776→VehicleID | Rank-5: 65.6 -> 66.8 | Improved |
| Use unified alignment for positive and negative sample pairs instead of separated alignment (ablation) | MSINet SAM · VeRi-776→VehicleID | Rank-5: 66.4 -> 65.8 | Hurt |
| Apply softmax to the activation values before spatial alignment (ablation) | SAM · VeRi-776→VehicleID | Rank-1: 49.0 -> 48.2 | Hurt |
| Add Spatial Alignment Module (ablation) | MSINet · MSMT17→Market-1501 | mAP: 46.2 -> 48.4 | Improved |
| Replace ResNet50 with MSINet as the backbone | GS unsupervised ReID · Market-1501 | mAP: 79.2 -> 81.5 | Improved |
| Replace ResNet50 with MSINet as the backbone | IDM unsupervised domain adaptation · Market-1501→MSMT17 | mAP: 33.5 -> 37.8 | Improved |
| Use a multiplication fusion operation instead of sum fusion (ablation) | MSINet multi-scale branches · MSMT17 | performance | Hurt |
| Use subtraction fusion instead of sum fusion (ablation) | MSINet multi-scale branches · MSMT17 | performance | No clear effect |
| Introduce receptive-field scale differences between the two branches (ablation) | MSINet MSI cells · MSMT17 | performance | Improved |
| Replace the searched interactions with a random interaction architecture (ablation) | MSINet MSI cells · MSMT17 | performance | Hurt |
Closest papers by meaning
- Signal: Selective Interaction and Global-local Alignment for Multi-Modal Object Re-Identification (AAAI 2026 · similarity 0.86)
- Hit-Detector: Hierarchical Trinity Architecture Search for Object Detection (CVPR 2020 · similarity 0.85)
- SM-NAS: Structural-to-Modular Neural Architecture Search for Object Detection (AAAI 2020 · similarity 0.84)
- Magic Tokens: Select Diverse Tokens for Multi-modal Object Re-Identification (CVPR 2024 · similarity 0.84)
- MambaPro: Multi-Modal Object Re-identification with Mamba Aggregation and Synergistic Prompt (AAAI 2025 · similarity 0.84)
- DC-NAS: Divide-and-Conquer Neural Architecture Search for Multi-Modal Classification (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.
