Research map / Vision-language and multimodal / Cross-modal learning
PMPGuard: Catching Pseudo-Matched Pairs in Remote Sensing Image–Text Retrieval
Pengxiang Ouyang, Qing Ma, Zheng Wang, Cong Bai
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 Cross-Gated Attention (CGA) (ablation) | PMPGuard-Geo · RSITMD (mismatch rate 0.8) | mR: 42.79 -> 49.36 | Improved |
| add Cross-Gated Attention (CGA) (ablation) | PMPGuard-Geo · RSICD (mismatch rate 0) | mR: 36.52 -> 40.23 | Improved |
| add Positive-Negative Awareness Attention (PNAA) (ablation) | PMPGuard-Geo · RSITMD (mismatch rate 0.8) | mR: 45.02 -> 49.36 | Improved |
| add Positive-Negative Awareness Attention (PNAA) (ablation) | PMPGuard-Geo · RSICD (mismatch rate 0) | mR: 38.56 -> 40.23 | Improved |
| use PMPGuard over DOVE under no added mismatch | PMPGuard · RSICD (mismatch rate 0) | mR: 23.79 -> 26.21 | Improved |
| use PMPGuard over DOVE under 0.8 mismatch | PMPGuard · RSITMD (mismatch rate 0.8) | mR: 28.68 -> 29.64 | Improved |
| use PMPGuard-Geo over HarMA-Geo at 0.8 mismatch | PMPGuard-Geo · RSITMD (mismatch rate 0.8, GeoRSCLIP backbone) | mR: 47.01 -> 49.36 | Improved |
| use PMPGuard-Geo over HarMA-Geo on real noisy data | PMPGuard-Geo · RS5M (GeoRSCLIP backbone) | mR: 32.78 -> 35.21 | Improved |
| increase synthetic mismatch rate from 0 to 0.8 | PMPGuard-Geo · RSICD (GeoRSCLIP backbone) | mR: 40.23 -> 36.54 | Hurt |
Closest papers by meaning
- Robust Remote Sensing Image-Text Retrieval with Noisy Correspondence (CVPR 2026 · similarity 0.90)
- TriSim: Tri-Dimensional Similarity Modeling with Extreme Value Theory for False-Negative Mitigation in Remote Sensing Image-Text Retrieval (CVPR 2026 · similarity 0.88)
- T-APT: Text-Guided Modality-Aware Prompt Tuning for Arbitrary Multimodal Remote Sensing Data Joint Classification (AAAI 2026 · similarity 0.86)
- CF-IPT: Cross-Modal Fusion Interactive Prompt Tuning of Vision-Language Pre-Trained Model for Multisource Remote Sensing Data Classification (CVPR 2026 · similarity 0.86)
- Cross-Modal and Uni-Modal Soft-Label Alignment for Image-Text Retrieval (AAAI 2024 · similarity 0.85)
- MMM-RS: A Multi-modal, Multi-GSD, Multi-scene Remote Sensing Dataset and Benchmark for Text-to-Image Generation (NeurIPS 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.
