A Novel Approach for Vehicular Accident Detection and Rescue Alert System using IoT with Convolutional Neural Network

Abstract

Road accidents are become a big concern for the public. This study describes a system for preventing accidents that involves alcohol detection with a MQ3 alcohol sensor, followed by a message alert to a rescue worker or family members. An SW-420 vibration sensor is used by the detecting component to identify any unusual vibrations that might result from a collision. Supervised deep learning CNN methods go along with this. The front camera of the car is utilized to take a picture of the accident scene for the deep learning accident prediction model. After a collision is discovered, communication is transmitted to the closest evacuation facility utilizing GPS and GSM modules. Once the Vehicle is involved in a collision following vehicle will be get notified through VANET. The alcohol sensor will then determine if the driver has consumed alcohol or not and if they need to operate the vehicle in an emergency when they are intoxicated. Then, as a result of driving too fast, multiple accidents occurred. So, when the car exceeds the speed limit, an immediate warning will be sent through GSM Module. Finally, Accident analysis device can be used for smart cities using supervised deep learning CNN methods.