SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection

2024-03-11

Neural Information Processing Systems

AI Summary

SAR target detection suffers from limited single-class datasets (typically <2,000 images) and scarce open-source implementations. To address this, we introduce SARDet-100K—the first COCO-scale, multi-class SAR benchmark comprising 100,000 images across 12 categories. We further propose MSFA, a multi-stage filtering augmentation pretraining framework that bridges the modality and structural gap between RGB pretraining and SAR fine-tuning via frequency-domain filtering enhancement, cross-domain feature alignment, and progressive fine-tuning. MSFA is detector-agnostic and seamlessly integrates with mainstream architectures including YOLO and Faster R-CNN. On SARDet-100K, it achieves over an 8.2% mAP improvement, demonstrating strong generalization and model independence. Both the full dataset and source code are publicly released to advance standardized, reproducible SAR detection research.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Synthetic Aperture Radar (SAR) object detection has gained significant attention recently due to its irreplaceable all-weather imaging capabilities. However, this research field suffers from both limited public datasets (mostly comprising<2K images with only mono-category objects) and inaccessible source code. To tackle these challenges, we establish a new benchmark dataset and an open-source method for large-scale SAR object detection. Our dataset, SARDet-100K, is a result of intense surveying, collecting, and standardizing 10 existing SAR detection datasets, providing a large-scale and diverse dataset for research purposes. To the best of our knowledge, SARDet-100K is the first COCO-level large-scale multi-class SAR object detection dataset ever created. With this high-quality dataset, we conducted comprehensive experiments and uncovered a crucial challenge in SAR object detection: the substantial disparities between the pretraining on RGB datasets and finetuning on SAR datasets in terms of both data domain and model structure. To bridge these gaps, we propose a novel Multi-Stage with Filter Augmentation (MSFA) pretraining framework that tackles the problems from the perspective of data input, domain transition, and model migration. The proposed MSFA method significantly enhances the performance of SAR object detection models while demonstrating exceptional generalizability and flexibility across diverse models. This work aims to pave the way for further advancements in SAR object detection. The dataset and code is available at https://github.com/zcablii/SARDet_100K.

Problem

• Lack of large-scale public SAR datasets for object detection
• Disparity between RGB pretraining and SAR finetuning challenges
• Need for open-source tools to advance SAR detection research

Innovation

• Established large-scale SARDet-100K benchmark dataset
• Proposed Multi-Stage Filter Augmentation pretraining framework
• Bridged RGB-SAR domain gaps via MSFA method