Gesture Recognition from Top-Down View
This project aims at recognizing human social interaction gestures in real time, using top-down view camera images obtained from cameras attached to the CCRL ceiling.
Project type:
Diploma Thesis
Project overview: From an overall of 40 cameras attached to the ceiling of a room, we obtain several continuous streams of images from a top-down perspective. The overall context of the project is to use these camera images to infer information about atomic actions of human social behaviour, i.e. gestures. More specifically, the focus of the this thesis is to recognize hand-shaking between two persons using the camera images and previously obtained tracking information about the position of persons in the scene.
Professor: Prof. Dr. Radig, Bernd, Ordinarius
Supervisor: Eggers, Martin
State: Running
Student: Michael Neumann
