Traffic Sign Detection Using Hybrid Network Of Yolo And Resnet

Abstract

Deep learning (DL) methods have outperformed machine learning (ML) and statistical techniques in predicting road traffic. Neural networks serve as the foundation for deep learning algorithms. In smart transport systems, the recognition and identification of traffic signs is a critical issue. Automated traffic sign detection and recognition Driving is essential for achieving autonomous transportation.. There are numerous traffic signs, and it takes time to train a good model. A TSR system can raise a driver’s awareness of the road situation and condition, potentially lowering traffic accidents. The YOLOv3 (You Only Look Once, Version 3) algorithm, a real-time object identification technique with cutting-edge features that detects individual items approach for detecting traffic flow, is what we suggest as a solution to this problem. The YOLO algorithm uses features that have been learned to find objects. For more accuracy we are combining two algorithms to make the prediction more efficient and accurate. ResNet is used to extract features from the training data set and the classification is done by YOLO V3 algorithm. Versions 1-3 of the YOLO machine learning algorithm were produced by the third version of the methodology, which is an improved version of the original DL algorithm. Before classification and the output of the optimised result, the image will undergo pre-processing to extract features. The proposed method makes it easier for drivers to detect traffic signs and reduces the number of accidents on the road. Experiment results show that the YOLO-v3 approach outperforms other previously existing algorithms in terms of average accuracy