To hide or seek facial recognition Thursday, 06 April 2017

News article written by Corbett Communications. The statements made or opinions expressed do not necessarily reflect the views of Engineers Australia.

Our face is our most basic form of identity, psychologically and physically, and its recognition is set to be the focus of future technology that will define us as individuals, just like the discovery of the fingerprint in the mid-19th century.

While facial recognition software has been used for some time for security purposes and detection of criminal behaviour, other forms of ID technology like implantable RFID chips are being trialled for ID access. However, the technology is being expanded for use in a range of areas from consumer purchases, personal safety, civil services, ticketed entry, and even endangered species monitoring.

Toowoomba City Council is currently conducting a trial of iOmniscient software – similar to that which the federal government uses for passport processing and the police use - as a pilot for major councils in Queensland. It works by analysing images of faces recorded by existing CCTV and is attached to a data system that will tell the council the number of people coming and going to the library and the number of times they actually come and go in any given day.

The old adage about a face that can unlock doors is about to become a literal reality, in particular in China. Baidu, the search engine and web services company, has teamed up with billion-dollar Beijing start-up Face++ to produce facial recognition software that is being used for access to tourist spots in the city of Wuzhen, instead of tickets. The director of Baidu’s Institute of Deep Learning, Yuangqing Lin, revealed that people’s facial recognition is processed “almost as quickly as walking by”.

While facial recognition is creeping into common use across society, it’s not totally infallible as researchers from Carnegie Melon University found late last year when they fooled state-of-the-art facial recognition software with multi-coloured glasses. But as the technology is perfected, we will see the creation of new apps and hardware as higher quality facial recognition becomes feasible even if a photo or video is grainy or shot at a strange angle, making it more difficult to hide from those who seek.

However, there are animals other than humans that need to be sought out if we’re going to help endangered species survive. The first-ever facial recognition system for lemurs, LemurFaceID, has been developed that can identify over a hundred different individuals with 98.7% accuracy, according to a paper published in BMC Zoology. The system was implemented using the OpenBR framework and was adapted from software designed to detect human faces, with a dataset of 462 different lemur faces of 80 individuals. The non-invasive technology has been found to remove many   limitations in monitoring endangered animals and has been recommended for long-term research as rapid, cost-effective and accurate.

Author: Desi Corbett


#facialrecognition #Baidu #deeplearning # Face++ #BMCZoology #OpenBR #LemurFaceID #RFID #iOmniscient