Traffic Sign Detection Algorithm Based on Improved YOLOv4

Abstract

Aiming at the problems of low detection accuracy of small targets, low real-time detection and missed target detection in the current traffic sign detection, an improved algorithm based on YOLOv4 is proposed. In the YOLOv4 network, the second residual block of CSPDarknet53 is used to embed the convolution structure and the YOLO head structure, increase another one output, and perform upsampling and downsampling in the network. The Inception structure is added to the classification and regression network to further improve the detection speed and increase the network complexity. Finally, the Inception structure is added to the classification and regression network to further improve the detection speed and increase the network complexity. The experimental results show that on the TT100K (Tsinghua-Tencent 100K) traffic sign dataset, compared with the YOLOv4 network, the mAP has increased by 3.6 percent, and the recognition speed has only decreased in a small range, basically achieving a balance between recognition accuracy and speed. The experimental results demonstrate the effectiveness of the improved network in small target detection and overall detection.