YOLOv8-based Spatial Target Part Recognition
Yongjing Zhou, Weigang Zhu, Yonghua He, Yonggang Li
Abstract
Inverse synthetic aperture radar (ISAR) images can reflect the finer structural features of space targets, which greatly promotes the development of feature extraction and recognition classification technology of space targets, and also becomes one of the important ways to monitor space targets and perceive the changes of space situation, which is of great significance to enhance our national defense strength. Significant significance. With the comprehensive development of machine learning, artificial intelligence, big data analysis and other multidisciplinary disciplines, the research on radar target characterization and recognition in deep learning has also made great progress. In this paper, we study a method based on YOLOvS (You Only Look Once) for the recognition of ISAR image components of spatial targets. Inspired by the idea of anchor-free, the algorithm discards the design of a priori frame and introduces the use of Task-Aligned Assigner positive and negative sample matching to improve the accuracy and speed of ISAR image target detection. The experimental results show that the detection method proposed in this paper can obtain better detection rate and quality factor. The detection rate of 98% was achieved in the tests conducted.