Object Detection and Tracking

PI: Yang Cai
Member: Guillaume Milcent
Contact : ycai@cmu.edu

Goals

After the bombing in London subways, investigators took four days to identify suspects of the bombers from 3,000 surveillance cameras in London subways.
 
The goal is to develop a semantic object detection system for real-time video processing as the event happens, as well as afterwards.

The by-product of this technology is the reduction of network data traffic and evidence retrieving time. Consequently, it would preserve privacy as the data emission becomes minimal, or on-demand.

Approach

Results

We tested algorithms on the real footages from Bombardier’s C-100 People Mover.

Applications

• Transit surveillance systems.
• Cargo security surveillance systems.
• Transportation safety systems.
• Passenger dynamics monitoring systems.
• Elderly care systems.
• Medical facilities.
• Border security systems.
• Onboard autonomous processing.
• Video on-demand transmission.
• Ambient intelligence systems.
• Smart TV systems.

Reference

Guillaume Milcent and Yang Cai, “Unattended baggage detection systems,” CyLab Technical Report, 2006.