引入CoordConv卷积的管制物品检测网络设计OA

The detection algorithm for controlled items with small targets has the problem of poor performance.To solve this problem,this paper proposes an improved YOLOv5 controlled item detection algorithm.Firstly,the CoordConv module is introduced into the network to represent the coordinates of feature map pixels.Then,the C2f module is introduced into the network to simultaneously utilize both detail and semantic information at different scales,improving the network's feature extraction ability and receptive field.The performance verification results of the algorithm show that compared to the YOLOv5 algorithm,our algorithm performs better on the Easy,Hard,and Hidden test sets mAP@.5.95 increased by 2.5,1.8,and 4.4 percentage points respectively,indicating that the detection accuracy of the algorithm in this paper is relatively high.