Driver Behavior Extraction from Videos in Naturalistic Driving Datasets with 3D ConvNets
This work provides a more efficient method for researchers to identify specific driver behaviors, like cell phone use, from large NDD video datasets, which is an incremental improvement for traffic safety and human factors research.
This paper addresses the challenge of underutilized video data in Naturalistic Driving Datasets (NDD) by developing a 3D ConvNet algorithm to automatically extract cell-phone-related driver behaviors. The method successfully extracts video chunks, with approximately 79% of them containing the targeted behaviors, significantly improving the efficiency of identifying these behaviors compared to manual review.
Naturalistic driving data (NDD) is an important source of information to understand crash causation and human factors and to further develop crash avoidance countermeasures. Videos recorded while driving are often included in such datasets. While there is often a large amount of video data in NDD, only a small portion of them can be annotated by human coders and used for research, which underuses all video data. In this paper, we explored a computer vision method to automatically extract the information we need from videos. More specifically, we developed a 3D ConvNet algorithm to automatically extract cell-phone-related behaviors from videos. The experiments show that our method can extract chunks from videos, most of which (~79%) contain the automatically labeled cell phone behaviors. In conjunction with human review of the extracted chunks, this approach can find cell-phone-related driver behaviors much more efficiently than simply viewing video.