Enhancing SAR Object Detection with Self-Supervised Pre-training on Masked Auto-Encoders

TL;DR

This paper introduces a self-supervised masked auto-encoder pre-training method tailored for SAR images, significantly enhancing object detection performance by learning domain-specific features without extensive labeled data.

Contribution

The paper proposes a novel SSL approach using Masked Auto-Encoders for SAR images, addressing the lack of domain-specific pre-training and improving detection accuracy.

Findings

Improves SAR object detection by 1.3 mAP on SARDet-100k benchmark.

Effectively captures latent SAR image representations.

Enhances model generalization in downstream tasks.

Abstract

Supervised fine-tuning methods (SFT) perform great efficiency on artificial intelligence interpretation in SAR images, leveraging the powerful representation knowledge from pre-training models. Due to the lack of domain-specific pre-trained backbones in SAR images, the traditional strategies are loading the foundation pre-train models of natural scenes such as ImageNet, whose characteristics of images are extremely different from SAR images. This may hinder the model performance on downstream tasks when adopting SFT on small-scale annotated SAR data. In this paper, an self-supervised learning (SSL) method of masked image modeling based on Masked Auto-Encoders (MAE) is proposed to learn feature representations of SAR images during the pre-training process and benefit the object detection task in SAR images of SFT. The evaluation experiments on the large-scale SAR object detection benchmark named SARDet-100k verify that the proposed method captures proper latent representations of SAR images and improves the model generalization in downstream tasks by converting the pre-trained domain from natural scenes to SAR images through SSL. The proposed method achieves an improvement of 1.3 mAP on the SARDet-100k benchmark compared to only the SFT strategies.