Head Pose Estimation Using a Chest-Mounted Camera and Its Evaluation Based on CG Images
Taisei Yamada, Hirotake Yamazoe · 2024
In everyday life, we naturally turn our heads or focus our gaze on objects of interest. By capturing a person's head pose or gaze direction, it is possible to record their actions and identify the objects they are focusing on. This study proposes a head pose estimation to infer and document the objects of interest based on head movements. We utilize a camera mounted on the user's chest and develop a deep-learning model to estimate head pose from the captured images. The training data was generated using a CG model, and the model's performance was validated.