SARMAE: Masked Autoencoder for SAR Representation Learning

TL;DR

SARMAE introduces a self-supervised learning framework for SAR imagery that leverages a large-scale dataset, speckle noise modeling, and optical priors to improve semantic understanding in various remote sensing tasks.

Contribution

The paper presents SARMAE, a novel masked autoencoder for SAR data that incorporates speckle noise awareness and optical prior alignment, enabling effective large-scale pre-training.

Findings

Achieves state-of-the-art results on SAR classification, detection, and segmentation.

Introduces SAR-1M, the first million-scale SAR dataset with paired optical images.

Demonstrates robustness and improved semantic representation in SAR tasks.

Abstract

Synthetic Aperture Radar (SAR) imagery plays a critical role in all-weather, day-and-night remote sensing applications. However, existing SAR-oriented deep learning is constrained by data scarcity, while the physically grounded speckle noise in SAR imagery further hampers fine-grained semantic representation learning. To address these challenges, we propose SARMAE, a Noise-Aware Masked Autoencoder for self-supervised SAR representation learning. Specifically, we construct SAR-1M, the first million-scale SAR dataset, with additional paired optical images, to enable large-scale pre-training. Building upon this, we design Speckle-Aware Representation Enhancement (SARE), which injects SAR-specific speckle noise into masked autoencoders to facilitate noise-aware and robust representation learning. Furthermore, we introduce Semantic Anchor Representation Constraint (SARC), which leverages paired optical priors to align SAR features and ensure semantic consistency. Extensive experiments across multiple SAR datasets demonstrate that SARMAE achieves state-of-the-art performance on classification, detection, and segmentation tasks. Code and models will be available at https://github.com/MiliLab/SARMAE.