Research map / Label-efficient and robust learning / Self-supervised and contrastive
SatMAE: Pre-training Transformers for Temporal and Multi-Spectral Satellite Imagery
Yezhen Cong, Samar Khanna, Chenlin Meng, Patrick Liu, Erik Rozi, Yutong He, Marshall Burke, David B. Lobell, Stefano Ermon
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 temporal encoding (ablation) | temporal SatMAE · fMoW RGB temporal | Top-1 accuracy: 78.07 -> 79.69 | Improved |
| Use independent masking instead of consistent masking (ablation) | temporal SatMAE · fMoW RGB temporal | Top-1 accuracy: 78.45 -> 79.69 | Improved |
| Add consistent cropping (ablation) | temporal SatMAE · fMoW RGB temporal | Top-1 accuracy: 79.90 -> 79.69 | Hurt |
| Add test-time augmentation (ablation) | temporal SatMAE · fMoW RGB temporal | Top-1 accuracy: 79.69 -> 81.49 | Improved |
| Stack temporal sequence along channels instead of temporal SatMAE (ablation) | SatMAE · fMoW RGB temporal | Top-1 accuracy: 81.49 -> 75.85 | Hurt |
| Group bands by wavelength/resolution (X) instead of random split (R) (ablation) | spectral SatMAE · fMoW Sentinel | Top-1 accuracy: 58.76 -> 59.30 | Improved |
| Group bands by wavelength (X) instead of halves (H) (ablation) | spectral SatMAE · fMoW Sentinel | Top-1 accuracy: 57.78 -> 59.30 | Improved |
| Use independent masking instead of consistent masking (ablation) | spectral SatMAE · fMoW Sentinel | Top-1 accuracy: 57.76 -> 59.30 | Improved |
| Add spectral group encoding (ablation) | spectral SatMAE · fMoW Sentinel | Top-1 accuracy: 58.87 -> 59.30 | Improved |
| Use ViT-Large instead of ViT-Base (ablation) | spectral SatMAE · fMoW Sentinel | Top-1 accuracy: 59.11 -> 59.30 | Improved |
| Lower masking ratio from 0.75 to 0.6 (ablation) | temporal SatMAE · fMoW RGB temporal | Top-1 accuracy: 7.19% | Hurt |
| Pre-train SatMAE+Group+IM vs scratch ResNet-152 | SatMAE+Group+IM · fMoW Sentinel | Top-1 accuracy: 49.12 -> 59.30 | Improved |
Later papers that compared with this method
4 later accepted papers report SatMAE as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- MOMO: Mars Orbital MOdel Foundation Model for Mars Orbital Applications (CVPR 2026 · better on 3 of 3 shared results)
- CROMA: Remote Sensing Representations with Contrastive Radar-Optical Masked Autoencoders (NeurIPS 2023 · better on 8 of 8 shared results)
- Scale-MAE: A Scale-Aware Masked Autoencoder for Multiscale Geospatial Representation Learning (ICCV 2023 · better on 7 of 7 shared results)
- Cross-Scale MAE: A Tale of Multiscale Exploitation in Remote Sensing (NeurIPS 2023 · better on 2 of 2 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
- Rethinking Transformers Pre-training for Multi-Spectral Satellite Imagery (CVPR 2024 · similarity 0.91)
- S2MAE: A Spatial-Spectral Pretraining Foundation Model for Spectral Remote Sensing Data (CVPR 2024 · similarity 0.88)
- DiffusionSat: A Generative Foundation Model for Satellite Imagery (ICLR 2024 · similarity 0.86)
- Accounting for Spatial Variability with the Histogram of Oriented Gradients Based Masking Improves Performance of Masked Autoencoder over Hyperspectral Satellite Imagery (Student Abstract) (AAAI 2025 · similarity 0.86)
- SkySense: A Multi-Modal Remote Sensing Foundation Model Towards Universal Interpretation for Earth Observation Imagery (CVPR 2024 · similarity 0.85)
- SatSynth: Augmenting Image-Mask Pairs through Diffusion Models for Aerial Semantic Segmentation (CVPR 2024 · 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.
