Data collection: Panoramic videos for motorcycling risk detection in Bangkok, Thailand, using Computer Vision

Keywords: Traffic accidents, Computer Vision, Motorcycle crashes
Abstract Type: Paper Abstract

Abstract

Thailand is one of the most dangerous countries for driving, especially for motorcyclists, with 20,000 deaths and over a million reported casualties per year. Yet, Thailand does not have official traffic accident records that contain factors associated with the crashes. Hence, it is challenging for researchers and practitioners to improve and evaluate the phenomena without a comprehensive dataset. However, there is a study, UCL’s 100 Cyclists Project, seeking to understand the factors affecting crash risk in cycling through the analysis of detailed near misses using Computer Vision algorithms to extract events from audio and video when a participant says ‘near miss’ to record the incident. This study, funded by Chulalongkorn University, will implement the former project by collecting a panoramic video dataset in a few streets from the Bangkok area's greatest density of historical motorcycling crashes. Motorcyclists will be given away a panoramic camera mounted to their helmets, and then they will identify risk scenarios by saying “an-ta-rai (Dangerous)”. The incidents relating to motorcyclists will be extracted and used to detail risk factors. The result can be used to identify the factors commonly associated with motorcycle crashes and adapt the results into policy to promote road accident prevention in Thailand and Southeast Asia. Decreasing the number of road fatalities as a new political declaration to halve road traffic deaths and injuries by 2030 is a milestone achievement.