Detection of Vehicle Crashes on Roads using Deep Learning

As the modern life is highly dependent on the modern transport system, road safety became a high priority. Most of the systems are aiming for very less reporting time to reduce the severity of the accidents. This paper presents an in-depth analysis of crash detection on roads using deep learning(DL). We have use one of the mechanisms of deepening i.e. Convolutional Neural Networks (CNNs) to detect the same. These DL models are used to classify the traffic accidents and it automatically recognize accidents with minimum effort. We have integrated automatic SMS alert with the system, which send SMS automatically after detecting the crash. This feature enhances the acceleration of emergency response, which makes the system more robust. We have achieved an accuracy of 96.3% which relatively good compared with the existing baseline methods.