Human Detection & Re-Identification for Mass Transit Environments
Large networks of cameras are ubiquitous in urban life, especially in densely populated environments such as airports, train stations, and sports arenas. For cost and practicality, most cameras in such networks are widely spaced so that their fields of view are non-overlapping. Automatically matching humans who re-appear across different cameras in such networks is a critical problem in homeland-security-related surveillance applications. This matching problem is closely related to a field of computer vision research called human re-identification, or “re-id” for short.
This project addresses the design and deployment of real-world re-id algorithms specifically designed for mass transit environments. This involves:
- the design and analysis of new computer vision algorithms for human detection, tracking, representation, and matching;
- the evaluation of the suitability of such algorithms for real-world homeland security applications, taking into account tracking/detection errors, latency/congestion, and human-computer interfaces to software systems; and
- the design of new experimental protocols and datasets that more closely resemble the types of re-id problems practitioners will encounter in real-world deployments.
The ideal end-state of the research is a suite of re-id algorithms that are directly applicable to the homeland security enterprise (HSE) and ready for large-scale system integration.Year 4 Annual Report
Rensselaer Polytechnic Institute
Students Currently Involved in Project
- Meng Zheng