一种改进YOLOv8n的电力设备红外图像识别网络OA

In view of the problems of low detection accuracy and large model calculation load in current infrared image recognition algorithms for power equipment,an improved infrared image recognition network for power equipment based on YOLOv8n,or YOLOv8n-DCSW is proposed.Firstly,in the YOLOv8n backbone network,Coordinate Attention(CA)is added and the standard convolution in the residual module is replaced with Deformable Convolution Network(DCN),which enhances the focus on small targets in complex environments and improves recognition accuracy.Secondly,the neck network is replaced with Sim-neck to reduce the computational complexity of the model.Finally,the Wise Intersection over Union(WIoU)loss function is introduced to reduce the gradient interference caused by low-quality borders and improve model recognition accuracy and convergence speed.Experimental results show that the proposed algorithm achieves a mean Average Precision(mAP)of 95.9％on the custom infrared dataset,with a computational cost of 6.9 GFLOPs.Compared to the original algorithm,the mAP has increased by 1.7％,while the computational cost has been reduced by 1.2 GFLOPs,meeting the requirements for high accuracy and low computation in the recognition of infrared images of electrical equipment.