Research map / Vision-language and multimodal / Cross-modal learning
AXM-Net: Implicit Cross-Modal Feature Alignment for 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 |
|---|---|---|---|
| Use implicit cross-modal AXM alignment and multi-scale features instead of the multi-scale joint-ID baseline | Person search · CUHK-PEDES test | Rank@1: 52.78 -> 62.08 | Helped |
| Add affinity loss (ablation) | AXM-Net + joint ID · CUHK-PEDES test | Rank@1: 62.08 -> 62.80 | Helped |
| Add triplet loss (ablation) | AXM-Net + joint ID · CUHK-PEDES test | Rank@1: 62.08 -> 63.72 | Helped |
| Add affinity loss to the triplet-loss model (ablation) | AXM-Net + joint ID + triplet · CUHK-PEDES test | Rank@1: 63.72 -> 64.44 | Helped |
| Add AXM-Block alignment (ablation) | Multi-scale joint-ID baseline · CUHK-PEDES test (Word2vec, 224 × 224) | Rank@1: 52.78 -> 55.90 | Helped |
| Add VFC (ablation) | Multi-scale joint-ID baseline · CUHK-PEDES test (Word2vec, 224 × 224) | Rank@1: 52.78 -> 56.99 | Helped |
| Add TSA (ablation) | Multi-scale joint-ID baseline · CUHK-PEDES test (Word2vec, 224 × 224) | Rank@1: 52.78 -> 53.25 | Helped |
| Add the global visual branch alongside VFC (ablation) | AXM-Net unified visual variant · CUHK-PEDES test (Word2vec, 224 × 224) | Rank@1: 54.53 -> 59.44 | Helped |
| Train in two stages instead of one stage (ablation) | AXM-Net · CUHK-PEDES test (Word2vec, 224 × 224) | Rank@1: 55.05 -> 59.44 | Helped |
| Share the VFC MLP across image strips rather than use separate MLPs (ablation) | AXM-Net VFC branch · CUHK-PEDES test | Rank@1: 57.62 -> 57.93 | Helped |
| Use max pooling rather than average pooling in VFC (ablation) | AXM-Net VFC branch · CUHK-PEDES test | Rank@1: 57.36 -> 58.82 | Helped |
| Apply random or part-based feature dropping (ablation) | AXM-Net VFC branch · CUHK-PEDES test | Rank@1: 59.44 -> 57.62; 59.44 -> 58.33 | Hurt |
Closest papers by meaning
- Learning Memory-Augmented Unidirectional Metrics for Cross-Modality Person Re-Identification (CVPR 2022 · similarity 0.90)
- Harmonious Attention Network for Person Re-Identification (CVPR 2018 · similarity 0.90)
- Implicit Discriminative Knowledge Learning for Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.89)
- Shallow-Deep Collaborative Learning for Unsupervised Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.89)
- Unifying Multi-Modal Uncertainty Modeling and Semantic Alignment for Text-to-Image Person Re-identification (AAAI 2024 · similarity 0.88)
- FlexiReID: Adaptive Mixture of Expert for Multi-Modal Person Re-Identification (ICML 2025 · 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.