Research map / Vision-language and multimodal / Cross-modal learning
CAST: Context-Aware Dynamic Latent Space Transformation for Interactive Text-to-Image Retrieval
Xuanzuo Lin, Min Zhang, Daizong Liu, Zhiwen Zuo, Xun Yang, Changting Lin, Xun Wang, Jianfeng Dong
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 context-aware low-rank projector (CLP) (ablation) | CAST on BLIP · VisDial (10-round avg) | Recall@1: 48.44 -> 51.04 | Improved |
| add context-guided modulator (CGM) on top of CLP (ablation) | CAST on BLIP · VisDial (10-round avg) | Recall@10: 80.90 -> 82.05 | Improved |
| use low-rank decomposed projection instead of full context-aware matrix W (ablation) | CLP · VisDial (10-round avg) | Recall@10: 80.57 -> 82.05 | Improved |
| make projector context-aware instead of a shared context-agnostic W (ablation) | CLP · VisDial (10-round avg) | Recall@10: 79.18 -> 82.05 | Improved |
| use one shared projector rather than separate context-agnostic projectors per round (ablation) | context-agnostic projection · VisDial (10-round avg) | Recall@10: 75.52 -> 79.18 | Improved |
| make modulation strength context-guided instead of a learnable scalar (ablation) | CGM · VisDial (10-round avg) | Recall@10: 81.51 -> 82.05 | Improved |
| add CASR to PlugIR and ChatIR | PlugIR, ChatIR · VisDial | Recall@K | Improved |
| add CASR to zero-shot BLIP | BLIP · VisDial | Recall@K | Improved |
| dynamic latent space versus static space across dialogue rounds | CAST vs ChatIR · VisDial Recall@1 | Recall@1: 42.78 -> 43.60 (round 1); 53.15 -> 57.56 (round 10) | Improved |
Closest papers by meaning
- Context-I2W: Mapping Images to Context-Dependent Words for Accurate Zero-Shot Composed Image Retrieval (AAAI 2024 · similarity 0.89)
- TIGeR: Unifying Text-to-Image Generation and Retrieval with Large Multimodal Models (ICLR 2025 · similarity 0.88)
- Beyond Global Similarity: Multi-Conditional Retrieval for Fine-Grained Cross-Modal Understanding (CVPR 2026 · similarity 0.88)
- ConText-CIR: Learning from Concepts in Text for Composed Image Retrieval (CVPR 2025 · similarity 0.87)
- MetaEmbed: Scaling Multimodal Retrieval at Test-Time with Flexible Late Interaction (ICLR 2026 · similarity 0.87)
- QuARI: Query Adaptive Retrieval Improvement (NeurIPS 2025 · 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.
