Accident detection from dashboard camera video
Earnest Paul Ijjina, Sanjay Kumar Sharma · 2019
Intelligent transport system plays a key role in the current digital era of smart cities. Due to increased population and number of vehicles, there is a proportionate increase in the number of accidents and their mortality rate. The video-based traffic monitoring system is one of the main approaches used to address this challenge but is limited to the region of surveillance. To overcome this limitation, we propose an approach for detecting accidents from dashboard camera video using computer vision-based techniques. The variation in visual information during accidents is analyzed to design a computationally less expensive accident detection model for practical use. As a result, accidents are identified from the change in visual information during an accident. It is evaluated on the new Dashboard Video Accident Detection (DVAD) dataset, proposed in this work. The experimental study suggests the effectiveness of the proposed approach for accident detection, that can be further extended to notify the concerned in case of an accident.