21 October 2018
More than HK$5 million has been spent developing the tracking software. Photos: HKU
More than HK$5 million has been spent developing the tracking software. Photos: HKU

3D HKU software makes suspects stand out from the crowd

A Hong Kong research team has developed 3D software that can help investigators fast forward through security camera footage to spot criminal suspects.

Ta Kung Pao reported Friday that the technology can cut the time taken to sift through an hour of tape from about six hours to as little as an hour and a half.

Professor Yu Yizhou, from the University of Hong Kong’s computer science department, said the software can also process multiple surveillance videos and create an easily understood 3D reenactment of a past event.

The technology synchronizes video from a large number of surveillance cameras to give investigators a global picture of a complex scenario.

The software can also help to identify and track targeted suspects, based on features including body gestures and the color and texture of their clothing and belongings, the report said.

Yu told the newspaper that various agencies, including customs and airport authorities, responded positively to the system when they were introduced to it last month. Other places like harbors, casinos and shopping malls, can also use the system to monitor crowds, he said.

The team behind the software is refining the technology and expects to launch it on the market in one or two years.

More than HK$5 million has been spent on the project, with most of the funding coming from the government’s Innovation and Technology Fund.

– Contact us at [email protected]


The technology can keep tabs on suspects and crowds using footage from various cameras.

Hong Kong researchers have software that can track suspects based the colour and texture of their clothing.

The software developed by HKU researchers can recreate an event using footage from multiple surveillance cameras.

EJ Insight reporter

EJI Weekly Newsletter

Please click here to unsubscribe