Re-Identification and Long-Term Anomaly Detection
Recognizing the same human as he or she moves through a network of cameras with non-overlapping fields of view is an important and challenging problem in security and surveillance applications. This is often called the re-identification or “re-id” problem. For example, in an airport security surveillance system, once a target has been identified in one camera by a user or program, we want to learn the appearance of the target and recognize him/her when he/she is observed by the other cameras. We call this type of re-id problem “tag-and-track”. In this project, we assess and refine real-time human detection and tracking algorithms to produce a set of person candidates (rectangles of pixels) in each frame of video. We then address the decision problem of whether given candidates correspond to the same person in videos from two different cameras.
We plan to additionally leverage the data aggregation enabled by a wide-area camera network in a mass transit environment to learn patterns of motion that are time-varying and investigate such patterns at very long time scales. For example, crowd behavior that is commonplace at 10 AM on a Monday may be highly unusual at 5 PM on a Sunday. The research challenge will be to extract such time-varying patterns of normalcy based on hundreds of hours of video and then to detect deviations that correspond to anomalous motion that can be sent to human operators for review.
Rensselaer Polytechnic Institute
Students Currently Involved in Project
- Matthew Reome
- Mahdee Jameel
- Srikrishna Karanam
- Austin Li