Research map / Label-efficient and robust learning / Self-supervised and contrastive
Change-Aware Sampling and Contrastive Learning for Satellite Images
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 improved geographical sampling (ablation) | SeCo · EuroSat | accuracy: 90.05 -> 91.42 | Helped |
| Add long-term temporal contrast (ablation) | SeCo with improved sampling · EuroSat | accuracy: 91.42 -> 92.56 | Helped |
| Add change-awareness to the contrastive loss (ablation) | SeCo with improved sampling and long-term contrast · EuroSat | accuracy: 92.56 -> 93.08 | Helped |
| Add improved geographical sampling (ablation) | SeCo · DynamicEarthNet | mIoU: 46.83 -> 47.24 | Helped |
| Add long-term temporal contrast (ablation) | SeCo with improved sampling · DynamicEarthNet | mIoU: 47.24 -> 49.68 | Helped |
| Add change-awareness to the contrastive loss (ablation) | SeCo with improved sampling and long-term contrast · DynamicEarthNet | mIoU: 49.68 -> 50.20 | Helped |
| Replace MoCo v2 with SimCLR while retaining CACo (ablation) | CACo · EuroSat | accuracy: 87.97 -> 90.83 | Helped |
| Replace MoCo v2 with SimCLR while retaining CACo (ablation) | CACo · OSCD | F1: 44.05 -> 47.50 | Helped |
| Use long-term feature distance instead of the normalized ratio estimate (ablation) | CACo change estimation · EuroSat | accuracy: 93.08 -> 91.17 | Hurt |
| Remove exponential moving average smoothing from the ratio estimate (ablation) | CACo change estimation · EuroSat | accuracy: 93.08 -> 91.98 | Hurt |
| Use seasonally aligned long-term pairs for change estimation (ablation) | CACo change estimation · EuroSat | accuracy: 93.08 -> 92.92 | Hurt |
| Use CACo rather than SeCo pretraining | ResNet-18 · OSCD | F1-score: 46.84 -> 50.29 | Helped |
Closest papers by meaning
- RSRCC: A Remote Sensing Regional Change Comprehension Benchmark Constructed via Retrieval-Augmented Best-of-𝑁 Ranking (NeurIPS 2026 · similarity 0.86)
- The Change You Want To Detect: Semantic Change Detection In Earth Observation With Hybrid Data Generationf (CVPR 2025 · similarity 0.85)
- Galileo: Learning Global & Local Features of Many Remote Sensing Modalities (ICML 2025 · similarity 0.85)
- Content-Adaptive Non-Local Convolution for Remote Sensing Pansharpening (CVPR 2024 · similarity 0.85)
- DiffusionSat: A Generative Foundation Model for Satellite Imagery (ICLR 2024 · similarity 0.85)
- FM-ChangeNet: Learning Change through Pathwise Feature Transport (NeurIPS 2026 · similarity 0.85)
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.