Personal re-identification innovation could extend camera tracking beyond facial biometrics

The utility of facial recognition in public surveillance systems could be significantly improved if the systems could continue to identify individuals whose faces cannot be analyzed, for reasons such as facial occlusion or inadequate image resolution. A method that researchers at Xi’an Jiaotong-Liverpool University call “personal re-identification” is intended to do just that, identifying people from visual data from cameras and ID card images to track individuals.

“Face recognition only works if the facial features are very clear which isn’t the case with many surveillance cameras,” XJTLU researcher Dr. Jimin Xiao told the University news service. Instead it is often the whole body that appears and the face is not clear due to low camera resolution. This is where re-identification comes in.”

Re-identification has an accuracy rate of about 90 percent so far, Xiao says, compared to more than 99 percent for facial recognition, but it is yet in the early stages of its development. The results of the team’s early testing are described in a paper recently published in the journal Pattern Recognition.

“If we can improve person re-identification technology, there are widespread benefits for public security,” Xiao says. “Take for example, an airport. What we are trying to develop is algorithms that help cameras scan the airport environment and identify when a known terrorist arrives.”

The key innovation is an Individual Aggregation Network, which the paper’s lead author, PhD student Yanchun Xie, says learns to minimize feature variations for the same person in different images. In real-world scenarios, those variations can be large, due to poor lighting, different poses, and other factors.

“The network incorporates a state-of-the-art object detection framework so that the camera automatically narrows in on the person in the image, such as a pedestrian crossing a road. This essentially isolates the person from its surroundings,” Xie says.

Xie also recently published a paper in the journal IEEE Transactions on Circuits and Systems for Video Technology on using reinforcement learning for visual object tracking.

Biometrics are being deployed to a growing number of CCTV networks, such as in Malaysia, and Brazilian authorities are considering deploying the technology to help deal with pervasive organized crime.

Article Topics

biometrics  |  cameras  |  facial recognition  |  identification  |  re-identification  |  surveillance