Pedestrian Orientation Estimation using CNN and Depth Camera 2020-01-0700
This work presents a method for estimating human body orientation using a combination of convolutional neural network (CNN) and stereo camera in real time. The approach uses the CNN model to predict certain human body key points then transforms these points into a 3D space using the stereo vision system to estimate the body orientation. The CNN module is trained to estimate the shoulders, the neck and the nose positions. Detecting of three points is required to confirm human detection and provides enough data to translate the points into 3D space. Human body orientation can be used in many applications as in robot-human interaction where the robot needs to know how to face the user or estimate the walking path of the user to avoid collision.