Research map / Vision-language and multimodal / CLIP and zero-shot
Frequency-Aware Vision-Language Multimodality Generalization Network for Remote Sensing Image Classification
Junjie Zhang, Feng Zhao, Hanqiang Liu, Jun Yu
This paper tried a common change: Test-time augmentation
add diffusion-based training-test-time augmentation (DTAug) · MU->TR · OA: 79.05 -> 85.70 Improved
What 84 accepted papers report for this change
- 75 improved
- 2 no clear effect
- 1 mixed
- 6 hurt
- BOLT: Decision‑Aligned Distillation and Budget-Aware Routing for Constrained Multimodal QA on Robots ICLR 2026 · add budget-aware test-time augmentation (bTTA) · Robo2VLM-1 · accuracy: 42.89 -> 50.50Improved
- Test-Time Perturbation Tuning with Delayed Feedback for Vision-Language-Action Models CVPR 2026 · increase test-time augmentation budget · LIBERO and Atari · performanceHurt
- MapPFN: Learning Causal Perturbation Maps in Context NeurIPS 2026 · add test-time augmentation over gene subsets · 100 genes · AUPRCImproved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
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 diffusion-based training-test-time augmentation (DTAug) (ablation) | FVMGN baseline · MU->TR | OA: 79.05 -> 85.70 | Improved |
| add DTAug (ablation) | FVMGN baseline · HU->TR | OA: 47.82 -> 73.92 | Improved |
| add multimodal wavelet disentanglement (MWDis) (ablation) | FVMGN · MU->TR | OA: 85.70 -> 88.55 | Improved |
| add spatial-frequency-aware image encoder (SFIE) (ablation) | FVMGN · MU->TR | OA: 88.55 -> 90.69 | Improved |
| add multiscale spatial-frequency alignment (MSFFA) (ablation) | FVMGN · MU->TR | OA: 90.69 -> 92.44 | Improved |
| add MSFFA (ablation) | FVMGN · HU->TR | OA: 80.33 -> 82.87 | Improved |
| use multiscale spatial+frequency alignment instead of frequency-only alignment (ablation) | FVMGN · MU->HU | OA: 86.36 -> 93.19 | Improved |
| use both shared and modality-specific texts instead of no text (ablation) | FVMGN · MU->TR | OA: 90.09 -> 92.44 | Improved |
| reduce training samples to 3% (ablation) | FVMGN · HU->MU | OA | Hurt |
Closest papers by meaning
- MMM-RS: A Multi-modal, Multi-GSD, Multi-scene Remote Sensing Dataset and Benchmark for Text-to-Image Generation (NeurIPS 2024 · similarity 0.87)
- DiffCLIP: Few-shot Language-driven Multimodal Classifier (AAAI 2025 · similarity 0.87)
- SkyScript: A Large and Semantically Diverse Vision-Language Dataset for Remote Sensing (AAAI 2024 · similarity 0.86)
- Remote Sensing Vision-Language Foundation Models without Annotations via Ground Remote Alignment (ICLR 2024 · similarity 0.86)
- Towards Open-Vocabulary Remote Sensing Image Semantic Segmentation (AAAI 2025 · similarity 0.86)
- GeoLLaVA-8K: Scaling Remote-Sensing Multimodal Large Language Models to 8K Resolution (NeurIPS 2025 · similarity 0.86)
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.
