Germany’s Fraunhofer Institute is full of clever little projects that genuinely break the barrier of where science is used today, and is one of the authors of the MP3 music format, and has come up with many codec breakthroughs and the like. Fraunhofer operates in many avenues of science not just digital media, and our eyes this week were drawn to a facial recognition system it seems to be working on, to tell if we are happy or sad.
The system takes a camera feed, and looks at all the faces in the shot (not just one) and by separating out the prominent facial elements like eyebrows, nose and mouth, and testing this against a pre-learned database of facial expressions, it can accurately classify the expression. Classifications such as happy, surprised, sad or angry are apparently possible.
Fraunhofer suggests that the work can be used to see how people react to advertising, decide how easy software instructions are, or help tune learning software, or be used to check the levels of concentration of car drivers.
There are of course major issues here. Civil liberty campaigners won’t panic yet, but at the first sign of something like this being introduced into daily usage, they might want assurances that the data isn’t going to be hijacked by the police, or some other agency that can behave in a Big Brother fashion.
The poor harassed car driver has enough to do with watching the road, keeping eyes on the speedometer to stop falling foul of speed cameras, but will now have the car beep at him every time he takes his eyes from straight ahead, and report him for dangerous driving every time he swerves.
And the idea of advertisements that watch you will increase further the number of cameras watching us from the moment we leave the doors of our homes, until the moment we can hide behind them once again. Nevertheless, the system sounds efficient enough to find a home in one application or another.