, , , , , , , , , , , , , , , , , , , , , ,


Interpreting thoughts based solely on a person’s facial expressions is not only intriguing but can be extremely practical in social interactions. The ability can be both instinctive as well as learned, but as is inevitable with all such ‘skills’ ever since Deep Blue beat Garry Kasparov, it leads to the question: can such a skill be taught to a computer?

Abhinav Dhall and his team of researchers at the Australian National University in Canberra believe they have the answer in the form of facial recognition software that can analyse a photograph, judge the face and give the picture an overall ‘mood score’.

Presented at the Conference on Multimedia Retrieval in Dallas, Texas, in April 2013, the face recognition system processes the expressions of the face by monitoring the positions of focus areas such as the corners of the mouth and eyes.

This software runs on an underlying machine learning algorithm, ‘trained’ on pre-labelled photographs, which it uses to rate the overall mood score of any new images.

The researchers tested their system against information from a control group which rated the intensity of an individual’s expression before comparing it to the overall mood score of a photo. The test results were encouraging – the system’s results were off compared to  human assessment by a mere seven %.

The possibilities of the system are immense, especially when considering its real-time advantages. According to Dhall, “By looking at a sequence of frames in a video, the system could even gauge the mood of a crowd in
real-time. If the mood score goes down, we can assume that the group is getting angry.”

Once the system is perfected, it is expected to find multiple arenas of use in security systems to complement lie detectors; in enhancing the video gaming experience by adjusting game elements according to a gamer’s mood; and in providing feedback to advertisers.

– Adi Abdurab

First published in the Adbuzzzz Section of The DAWN National Weekend Advertiser on June 9, 2013.