The latest DNN-based developments in people tracking such as OpenPose dramatically changed the people tracking and analysis capabilities. This project will use OpenPose as basic detection brick to analyse crowds. The analysis includes people long-term tracking/reidentification (in a closed dataset) and the extraction of groups of people from the crowd. The results will be mapped into a simple augmented reality framework to demonstrate its effectiveness.

Project goals

The goal of the project is to analyse crowds based on video input and extract individual people as well as groups of people.

Project leader

Matei Mancas (matei.mancas@gmail.com)

Project organisation

Work package 1 : people tracking
Work package 2 : extracting groups from crowds