AI-Safe Transportation: Real-Time Incident Detection and Alerting System in Smart Cities

Recently, with the increasing complexity of urban environments, the need for advanced technologies to ensure safe transportation in smart cities has become paramount. It has become necessary also to propose a model based on the latest technology, which can detect various road and environmental accidents in real time using live feed from street cameras. In this work, the proposed model can identify incidents such as car accidents, fires, floods, and other malfunctions in the road environment, and automatically generates two reports - one for the government to seek assistance and another for drivers to avoid the affected area and mitigate traffic congestion. The developed model utilizes state-of-the-art Deep learning techniques. It is implemented by convolutional neural networks (CNNs), for efficient processing and analysis of live video feeds. The model is trained on a comprehensive dataset of images, encompassing different types of road and environmental incidents. The trained model is subsequently integrated into a real-time monitoring system that continuously receives and analyzes live feeds from multiple cameras installed in the streets. The main features of the developed system include real-time detection and classification of road and environmental incidents, automatic generation of reports for government and users, and proactive traffic management to reduce traffic congestion in affected areas. The system is designed to work seamlessly with existing smart city infrastructure and can be easily integrated into transportation management systems. The experimental results proved that the deep learning YOLO model with a dataset licensed by MIT achieved an average accuracy of 91% in detecting various types of road and environmental incidents in real-time.