Activity: Talk or presentation types › Oral presentation
Description
In this talk, we will show how the use of modern machine learning, video analytics and AI can be used to assess and monitor crowd behaviour in public spaces. In our system, people of other relevant object of interest are first detected and tracked using deep learning and tracking algorithms over a full network of non-overlapping cameras. This crucial but sometime noise information is then used to feed an event recognition engine able to model and manage uncertainty. This allows us to compose complex semantic events happening in our scenario and trigger relevant alarms, as well as filter false positive to avoid user disengagement. In this talk, this technology will be demonstrated for crowd analysis in public transport hubs.