NUDT4MSTAR
收藏资源简介:
NUDT4MSTAR是由国防科技大学创建的一个大规模合成孔径雷达(SAR)数据集,旨在推动SAR自动目标识别(ATR)技术的发展。该数据集包含194,324张图像,涵盖了40种车辆目标类型和5种不同场景的成像条件,是现有同类数据集中规模最大的。数据集不仅提供了处理后的幅度图像,还包含了原始的复数格式数据,并详细标注了每张图像的目标信息和成像条件。通过构建包含7个实验和15种识别方法的基准测试,NUDT4MSTAR展示了其在SAR目标识别领域的广泛应用潜力,尤其是在复杂场景下的目标识别问题。该数据集的开源将有助于推动SAR ATR技术的进一步发展,并吸引更多研究者的关注。
NUDT4MSTAR is a large-scale Synthetic Aperture Radar (SAR) dataset developed by the National University of Defense Technology, aiming to advance the development of SAR Automatic Target Recognition (ATR) technologies. Comprising 194,324 images, this dataset covers 40 types of vehicle targets and 5 different imaging conditions across scenarios, making it the largest dataset of its kind currently available. The dataset provides not only processed amplitude images but also raw complex-format data, with detailed annotations of target information and imaging conditions for each image. By establishing a benchmark test including 7 experiments and 15 recognition methods, NUDT4MSTAR demonstrates its broad application potential in the field of SAR target recognition, particularly for target recognition tasks in complex scenarios. The open release of this dataset will help further advance SAR ATR technologies and attract more researchers' attention.

- 1NUDT4MSTAR: A New Dataset and Benchmark Towards SAR Target Recognition in the Wild国防科技大学 · 2025年



