Research map / Vision-language and multimodal / Cross-modal learning
INTENT: Invariance and Discrimination-aware Noise Mitigation for Robust Composed Image Retrieval
Zhiwei Chen, Yupeng Hu, Zhiheng Fu, Zixu Li, Jiale Huang, Qinlei Huang, Yinwei Wei
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 Visual Invariant Composition (FFT counterfactual consistency) (ablation) | INTENT · FashionIQ-Avg R@10, 20% noise | R@10: 53.05 -> 54.22 | Improved |
| use FFT counterfactual images instead of grayscale images (ablation) | INTENT · FashionIQ-Avg R@10, 20% noise | R@10: 54.01 -> 54.22 | Improved |
| add positive weight reward to decision boundary (ablation) | INTENT · FashionIQ-Avg R@10, 20% noise | R@10: 53.37 -> 54.22 | Improved |
| add negative weight reward to decision boundary (ablation) | INTENT · FashionIQ-Avg R@10, 20% noise | R@10: 53.89 -> 54.22 | Improved |
| add both rewards (scalable decision boundary) (ablation) | INTENT · CIRR-Avg R@K, 20% noise | R@K: 77.93 -> 80.17 | Improved |
| add robust contrastive loss (ablation) | INTENT · CIRR-Avg R@K, 20% noise | R@K: 78.33 -> 80.17 | Improved |
| mask positives in the robust contrastive loss (ablation) | INTENT · CIRR-Avg R@K, 20% noise | R@K: 74.26 -> 80.17 | Improved |
| add soft discriminative loss (ablation) | INTENT · CIRR-Avg R@K, 20% noise | R@K: 77.67 -> 80.17 | Improved |
| use CKA instead of MSE for causal consistency loss (ablation) | INTENT · FashionIQ-Avg R@10, 20% noise | R@10: 53.88 -> 54.22 | Improved |
| INTENT vs TME under high noise | INTENT · FashionIQ val, 80% noise | AVG: 57.63 -> 59.07 | Improved |
| INTENT vs TME under no noise | INTENT · CIRR test, 0% noise | Avg(R@5, Rsub@1): 82.01 -> 81.70 | No clear effect |
| INTENT vs TME at 20% noise | INTENT · CIRR test, 20% noise | Avg(R@5, Rsub@1): 79.74 -> 79.66 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports INTENT as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Air-Know: Arbiter-Calibrated Knowledge-Internalizing Robust Network for Composed Image Retrieval (CVPR 2026 · better on 3 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning with Noisy Triplet Correspondence for Composed Image Retrieval (CVPR 2025 · similarity 0.93)
- Do Composed Image Retrieval Benchmarks Require Multimodal Composition? (NeurIPS 2026 · similarity 0.90)
- Instance-Level Composed Image Retrieval (NeurIPS 2025 · similarity 0.90)
- Data Roaming and Quality Assessment for Composed Image Retrieval (AAAI 2024 · similarity 0.88)
- Vision-by-Language for Training-Free Compositional Image Retrieval (ICLR 2024 · similarity 0.88)
- Aligning Composed Query with Image via Discriminative Perception from Negative Correspondences (AAAI 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.
