Vision-based Tracking of Multiple Persons in an Emergency Room

Thesis by: n.n.
Advisor: Prof. Nassir Navab, Dr. Marc Lazarovici
Supervision by: Diana Mateus, Loren Schwarz

Abstract

The Institute for Emergency Medicine and Management in Medicine (INM) offers facilities to perform professional trainings for emergency personnel. A fully equipped emergency room allows to simulate various critical situations and to record the performance of emergency personnel by means of multiple, synchronized modalities, such as digital patient data streams, video cameras, and microphones. This information is used afterwards by a trainer to evaluate aspects of the performance together with the trainees.

Integrating positional information for individual persons in the synchronized recording system would first allow to facilitate evaluation. While currently the movements of persons within the emergency room have to be followed in recorded video images, a bird's eye per-spective visualization would help to keep an overview of the situation. Second, additional insight into the quality of the training performance could be gained by studying the relative trajectories of persons, thus gaining insight about personal and team interactions. Also, new training options in defining standardized positions for different situations can be developed. Such a tracking system for persons can furthermore be a starting point for various other ap-plications, such as automatic activity recognition.

Project Outline

The goal of this Master's Thesis is to implement a vision-based tracking system that allows to follow the position of multiple persons within a room. The desired output is 2D-positional information for each person with respect to the room coordinate frame that can be visualized in the existing interactive virtual floor plan. Video cameras as present at the INM are to be used as an input.

In a first stage, tracking can be performed by the help of markers attached to the persons. In a later stage, natural features, e.g. color and shape, could also be used. The cameras have to be calibrated with respect to the common ground plane and tracked feature positions need to be correctly projected onto the ground plane. Challenges include the typical lighting conditions in an emergency room, distortions introduced by wide angle video cameras, as well as visual occlusions due to machinery and different persons. The final system should there-fore also comprise a simple model-based filtering approach that allows to cope with these issues.

Application

Interested students with strong abilities in C++ programming and knowledge of 3D computer vision are highly encouraged to contact Loren Schwarz or Diana Mateus for more information.