Hybrid Approach for Detecting the Traffic Violations Based on Deep Learning Using the Real-Time Data

Nowadays, the number of vehicles on the road is steadily increasing, and this raises the risk of road accidents. One of the reasons of this regrettable situation is violation of traffic rules. Overloaded cars are one example of a traffic rule violation. This research study proposes a hybrid approach for detecting different types of violations and violators. The proposed model will detect the overloaded vehicles and the number plate of the offenders by using the deep neural network model called YOLOv4. Additionally, Tesseract is used to recognize characters in the LSTM number plate. Finally, the proposed model has delivered a satisfactory result.