基于YOLOv5的小动物目标检测算法研究
Research on Small Animal Target Detection Algorithm Based on YOLOv5
汪香念 1饶红霞 1谢家豪1
作者信息
- 1. 广东工业大学自动化学院,广东 广州 510006
- 折叠
摘要
针对现有变电站入侵检测算法误报率高、对小目标检测精度低等问题,提出了一种基于改进YOLOv5 的变电站入侵小动物目标检测算法.将SENet通道注意力模块和卷积注意力模块(CBAM)中的激活函数改进为HardSwish函数,并在主干网络和颈部网络中分别引入改进后的SENet_H模块和CBAM_H模块;采用空洞空间池化金字塔(ASPP)对空间金字塔池化进行优化,并在检测端增加一个小目标检测层,以提高对小动物的检测精度.此外,还构建了小动物数据集,并采用9-Mosaic数据增强方式,丰富了样本目标.实验结果表明:改进后的小动物目标检测算法相较于原YOLOv5 算法精确率提升了 11.6%,召回率提升了 10.2%,平均精度均值提升了 8.1%.
Abstract
Aiming at the problems of high false alarm rate and low detection accuracy of small targets in existing sub-station intrusion detection algorithms,a substation intrusion small animal target detection algorithm based on improved YOLOv5 is proposed.The activation function in the SENet channel attention module and Convolutional Block Attention Mod-ule(CBAM)is improved to HardSwish function,and the improved SENet_H module and CBAM_H module is separately intro-duced into the backbone network and the neck network.Experimental results show that the improved small animal target detection algorithm outperforms the original YOLOv5 algorithm by 11.6%in precision,10.2%in recall,and 8.1%in mean average precision.
关键词
目标检测/注意力机制/空洞空间池化金字塔/小动物检测Key words
target detection/attention mechanism/atrous spatial pyramid pooling/small animal detection引用本文复制引用
出版年
2024
NETL
NSTL
万方数据