Traffic sign detection based on depth improved YOLOV5
Zhaoyang Xie, Taijun Li · 2023
Traffic sign detection is an important part of the intelligent transportation system, which is of great significance to assisted driving, unmanned driving and intelligent driving. The sign detection algorithm needs to be lightweight. In order to solve these problems, reduce the amount of network calculations and improve the detection speed of the network without affecting the accuracy of the algorithm, this paper proposes an algorithm using the lightweight network model Ghost net as the backbone network. By embedding the CBAM attention mechanism in the network, the influence of interference information is reduced, and the feature extraction ability of the network is improved. At the same time, the weight-removed BIFPN module is introduced to strengthen the feature fusion ability of the network and improve the detection effect of small targets. Experiments show that the improved algorithm has achieved an excellent level of 80.2% in the Map_0.5 performance index on the CCTSDB China traffic sign detection dataset, and the FPS index has increased by 37% compared with the classic algorithm.