Media

Professor Yu Yizhou (left) and Dr Calvin Fong (right) 

The software processes surveillance videos and creates an informative and easy-to-comprehend 3D reenactment of a past event to assist an incident or crime investigation. Such a reenactment, which can synchronize video footages from a large number of surveillance cameras, lets investigators have a global situational understanding of a complex scenario without the need to loop through individual cameras, and provides useful reference for situations including crime scene reconstruction. The software can further identify and track targeted figures in the videos, using features including body gestures, colour and texture of clothing and belongings.

Unlike existing video analytic software, the one developed by the HKU team is much more advanced in terms of three dimensional reconstruction and visualization capabilities, which allow close-up examination of the actions of suspects from all possible viewing angles.

The new software can save time and manpower in investigations. The conventional way of collecting evidence from surveillance videos requires investigators to loop through images retrieved from many cameras. They have to view the footages again and again so as to have an understanding of the crime scene and identify suspects. This process can be very labour and resource intensive. The new software can help track the suspects and their collaborators in a systematic manner, to know their whereabouts under different time frames, and retrieve subtle but important details which could easily be overlooked otherwise.

Places that can benefit from this system include airports, harbours, casinos, shopping malls and busy city blocks. It can be a useful tool to be used by government agencies including the police force, the immigration department, and port and airport authorities, for investigation and terrorist control purposes.

Dr Calvin Fong, member of the research team, said: “The new software is like a ‘Smart 3D SkyEye’ which integrates images from dozens of or even more surveillance cameras. It should be a lot more convenient to identify and track targeted figures using this new software than the traditional way of investigation.”

Professor Yu Yizhou, head of the project, said: “The software provides a more systematic and effective way of utilizing CCTV videos in investigations. It can be further developed for use in crime prevention, by integrating more advanced features such as body proportions. It can also be applied in crowd monitoring and group behavior tracking.”

For an animation video of the software, please visit: 
http://www.waterman.hku.hk/temp/Visualization-Steal.mp4

For media enquiries please contact Ms Melanie Wan, Communications and Public Affairs Office, HKU (Tel: 2859 2600 / Email: melwkwan@hku.hk); or Ms Jolly Li, Faculty of Engineering, HKU (Tel: 2241 5953 / E-mail: jollyli@hku.hk).