Kinect-Supported Dataset Creation for Human Pose Estimation

Dataset creation of annotated human poses is very time-consuming. This paper presents an approach for rapid construction of a precisely annotated training dataset for human pose estimation of a sitting subject, intended especially for aeronautic cockpit. We propose to use Kinect as a tool for collecting ground truth to a purely visual dataset (for reasons defined by the application, use of Kinect or similar structured light-based approaches is impossible).  Since Kinect annotation of individual joints might be imprecise at certain moments, manual post-processing of the acquired data is necessary and we propose a scheme for efficient and reliable manual post-annotation. We produced a dataset of 6,322 annotated frames, involving 11 human subjects recorded in various light condition, different clothing, and varying background. Each frame contains one seated person in frontal view with annotation of pose and optical flow data.