Papers › Feature Guided Masked Autoencoder for Self-supervised Learning in Remote Sensing

Feature Guided Masked Autoencoder for Self-supervised Learning in Remote Sensing

28 Oct 2023arXiv:2310.18653archive 2025-07-28

Yi Wang, Hugo Hernández Hernández, Conrad M Albrecht, Xiao Xiang Zhu

Self-supervised learning guided by masked image modelling, such as Masked AutoEncoder (MAE), has attracted wide attention for pretraining vision transformers in remote sensing. However, MAE tends to excessively focus on pixel details, thereby limiting the model's capacity for semantic understanding, in particular for noisy SAR images. In this paper, we explore spectral and spatial remote sensing image features as improved MAE-reconstruction targets. We first conduct a study on reconstructing various image features, all performing comparably well or better than raw pixels. Based on such observations, we propose Feature Guided Masked Autoencoder (FG-MAE): reconstructing a combination of Histograms of Oriented Graidents (HOG) and Normalized Difference Indices (NDI) for multispectral images, and reconstructing HOG for SAR images. Experimental results on three downstream tasks illustrate the effectiveness of FG-MAE with a particular boost for SAR imagery. Furthermore, we demonstrate the well-inherited scalability of FG-MAE and release a first series of pretrained vision transformers for medium resolution SAR and multispectral images.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

zhu-xlab/fgmae officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image ClassificationMulti-Label Image ClassificationSelf-Supervised Learning

Datasets

Introduced by this paper, per the archive.

EuroSAT-SAR

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification EuroSAT-SAR FG-MAE (ViT-S/16) Overall Accuracy 85.9 #1 of 3 Archive leaderboard report
Image Classification EuroSAT-SAR MAE (ViT-S/16) Overall Accuracy 81.0 #2 of 3 Archive leaderboard report
Image Classification EuroSAT-SAR ViT-S/16 Overall Accuracy 78.4 #3 of 3 Archive leaderboard report
Multi-Label Image Classification BigEarthNet (official test set) FG-MAE (ViT-S/16) F1 Score 80.8 #1 of 6 Archive leaderboard report
Multi-Label Image Classification BigEarthNet (official test set) FG-MAE (ViT-S/16) mAP (micro) 89.3 #1 of 6 Archive leaderboard report
Multi-Label Image Classification BigEarthNet (official test set) MAE (ViT-S/16) F1 Score 79.9 #4 of 6 Archive leaderboard report
Multi-Label Image Classification BigEarthNet (official test set) MAE (ViT-S/16) mAP (micro) 88.6 #4 of 6 Archive leaderboard report
Multi-Label Image Classification BigEarthNet (official test set) ViT-S/16 F1 Score 78.9 #5 of 6 Archive leaderboard report
Multi-Label Image Classification BigEarthNet (official test set) ViT-S/16 mAP (micro) 87.8 #5 of 6 Archive leaderboard report
Multi-Label Image Classification BigEarthNet-S1 (official test set) FG-MAE (ViT-S/16) mAP (micro) 82.7 #1 of 3 Archive leaderboard report
Multi-Label Image Classification BigEarthNet-S1 (official test set) MAE (ViT-S/16) mAP (micro) 81.3 #2 of 3 Archive leaderboard report
Multi-Label Image Classification BigEarthNet-S1 (official test set) ViT-S/16 mAP (micro) 79.5 #3 of 3 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

FocusMAE

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections