Traffic Violation Prediction Using Deep Learning

Abstract

Smart transportation infrastructure is needed and technology is now ready to help us. Artificial intelligence (especially deep learning) can help improve the performance of existing systems in many ways. The ability to effectively identify and classify tools is critical to the success of an intelligent machine. This device can play an important role in helping us travel in a country like India where there are many people but limited space. We present an advanced framework for detecting cyclists ignoring helmet rules using deep convolutional neural networks (CNNs). The system includes motorcycle, detection, helmet and no helmet, classification and calculation using the YOLO algorithm. Convolutional Neural Network with Sequential CNN model is used in the license control process. The CNN classification model proposes to classify the permission in the image and extract the user's content. Then calculate the penalty. Finally, make an SMS service that sends reminders to users to avoid motorcycle accidents and pay within a week. Automatically block accounts and send notifications if user doesn't pay. Finally, the administrator updates the user's license after receiving the penalty. We measure the accuracy and speed of the frame.