摘要
针对无人机遥感影像中的车辆检测问题,本文对YOLOv5算法中损失函数仅利用边界框长宽比不能反映边界框宽高分别与其置信度的真实差异的缺点,对损失函数进行了改进,消除了因为不能反映宽高真实差异引起的收敛过慢问题。利用改进的YOLOv5算法进行了训练和检测,验证了该算法对于无人机航拍影像中车辆检测的有效性。
Aiming at the problem of vehicle detection in UAV remote sensing images, this paper improves the loss function of YOLOv5 algorithm, which only uses the length-width ratio of the bounding box and cannot reflect the real difference between the width and height of the bounding box and its confidence. It eliminates the problem of slow convergence caused by the failure to reflect the real difference in width and height, and uses the improved YOLOv5 algorithm for training and detection. The effectiveness of the algorithm for vehicle detection in UAV aerial images is verified.
出处
《人工智能与机器人研究》
2022年第4期325-332,共8页
Artificial Intelligence and Robotics Research
