Research map / Vision-language and multimodal / Cross-modal learning
Identity-Aware Textual-Visual Matching With Latent Co-Attention
Shuang Li, Tong Xiao, Hongsheng Li, Wei Yang, Xiaogang Wang
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 |
|---|---|---|---|
| use CMCE loss instead of triplet loss (ablation) | Stage-1 · CUHK-PEDES text-to-image retrieval | top-1 accuracy: 14.76 -> 21.55 | Improved |
| use CMCE loss instead of triplet loss (ablation) | Stage-1 · CUB image-to-text | top-1 accuracy: 52.5 -> 61.5 | Improved |
| use CMCE loss instead of triplet loss (ablation) | Stage-1 · CUB text-to-image | AP@50: 52.4 -> 55.5 | Improved |
| use CMCE loss instead of triplet loss (ablation) | Stage-1 · Flower image-to-text | top-1 accuracy: 64.3 -> 68.4 | Improved |
| use CMCE loss instead of triplet loss (ablation) | Stage-1 · Flower text-to-image | AP@50: 64.9 -> 68.0 | Improved |
| use identity-level annotations (ablation) | Stage-2 · CUHK-PEDES text-to-image retrieval | top-1 accuracy: 23.47 -> 25.94 | Improved |
| use identity-level annotations (ablation) | Stage-2 · CUHK-PEDES text-to-image retrieval | top-10 accuracy: 54.77 -> 60.48 | Improved |
| add latent semantic attention (ablation) | Stage-2 · CUHK-PEDES text-to-image retrieval | top-1 accuracy: 23.58 -> 25.94 | Improved |
| add latent semantic attention (ablation) | Stage-2 · CUHK-PEDES text-to-image retrieval | top-10 accuracy: 58.68 -> 60.48 | Improved |
| add spatial attention (ablation) | Stage-2 w/o SMA · CUHK-PEDES text-to-image retrieval | top-1 accuracy: 22.11 -> 23.58 | Improved |
| add spatial attention (ablation) | Stage-2 w/o SMA · CUHK-PEDES text-to-image retrieval | top-10 accuracy: 58.05 -> 58.68 | Improved |
| add stage-1 initialization and easy-negative screening (ablation) | Stage-2 w/o SMA+SPA · CUHK-PEDES text-to-image retrieval | top-1 accuracy: 17.19 -> 22.11 | Improved |
Closest papers by meaning
- Multi-Modality Cross Attention Network for Image and Sentence Matching (CVPR 2020 · similarity 0.90)
- Instance-Aware Image and Sentence Matching With Selective Multimodal LSTM (CVPR 2017 · similarity 0.89)
- Expressing Objects Just Like Words: Recurrent Visual Embedding for Image-Text Matching (AAAI 2020 · similarity 0.89)
- Image and Sentence Matching via Semantic Concepts and Order Learning (TPAMI 2018 · similarity 0.89)
- Text Embedding is Not All You Need: Attention Control for Text-to-Image Semantic Alignment with Text Self-Attention Maps (CVPR 2025 · similarity 0.88)
- Composing Object Relations and Attributes for Image-Text Matching (CVPR 2024 · 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.
