Driver Pose Estimation Using a Mixture-model Method
Gang Liu, Xinping Yan, Yufen Sun · 2013
Estimating driver pose from videos is a significant task in driver assistance systems. In this paper, we introduced a mixture model for driver pose estimation. By the mixture model, a driver's upper body is composed of a set of non-oriented parts, and the spatial relationships of parts are modeled by a collection of springs. We described the algorithm of estimation. The steps of the algorithm, such as calculating HOG features, distance transformation, message passing, and back tracking, are discussed in details. The experiment demonstrates that our implementation can achieve good performance.