Automatic Detection of Traffic Accidents from Video Using Deep Learning Techniques

• DOI: 10.3390/computers10110148
• Publication Date: 2021-11-09T17:46:23Z
• Authors: Sergio Robles-Ser...; German Sanchez-To...; John Branch-Bedoya
• Keywords: Identification

According to worldwide statistics, traffic accidents are the cause of a high percentage violent deaths. The time taken send medical response accident site is largely affected by human factor and correlates with survival probability. Due this wide use video surveillance intelligent systems, an automated detection approach becomes desirable for computer vision researchers. Nowadays, Deep Learning (DL)-based approaches have shown performance in tasks that involve complex features relationship. Therefore, work develops DL-based method capable detecting on video. proposed assumes events described visual occurring through temporal way. extraction phase, followed temporary pattern identification, compose model architecture. learned training phase convolution recurrent layers using built-from-scratch public datasets. An accuracy 98% achieved datasets, showing capacity independent road structure.