Survey on In-vehicle Datasets for Human Pose Estimation

Computer vision techniques are essential in autonomous driving applications. Consequently, many deep learning-based methods have been developed using a large number of outdoor images captured from vehicle-mounted cameras. However, less attention has been paid to the inside of vehicles, although in-vehicle applications such as driver monitoring and action recognition are also essential in the autonomous driving scenario. In this paper, we investigate the existing in-vehicle datasets, especially on human pose estimation, and provide their details.