火力与指挥控制2023，Vol.48Issue(11)：58-66,9.DOI:10.3969/j.issn.1002-0640.2023.11.009
改进的YOLOv5地面军事目标识别算法
Improved YOLOv5 Ground Military Target Recognition Algorithm
摘要
Abstract
Aiming at the problem of low performance of ground military target recognition algorithm in complex battlefield background,a PAL-YOLO ground military target recognition algorithm based on parallel attention mechanism is proposed.The algorithm re-clusters the anchor boxes of the target under the self-built ground military target data set;adds a channel-space parallel attention mechanism module to the Backbone of the network to improve the extraction ability of target feature;uses Alpha_IoU to improve the loss function of the target recognition classifier and to speed up the model convergence.The results show that the improved algorithm can improve the mAP value by 6.4％and the FPS by 6％while ensuring the model space complexity.关键词
目标识别/PAL-YOLO/注意力机制/Alpha-IoUKey words
target recognition/PAL-YOLO/attention mechanism/Alpha-IoU分类
信息技术与安全科学引用本文复制引用
刘康,宋晓茹,高嵩,陈超波..改进的YOLOv5地面军事目标识别算法[J].火力与指挥控制,2023,48(11):58-66,9.基金项目
机电动态控制重点实验室开放课题基金(6142601200301) （6142601200301）
陕西省重点研发计划项目(2021GY-287) （2021GY-287）
