The Eye Tribe

dear all..

Is it possible to measure eyelid closure using this device?

hatma

I was wondering, is it possible to develop a drowsiness and distraction monitoring system ( http://www.mynewsdesk.com/us/tobii_tech ... ems-755160 ) using theeyetribe eye tracking device?

It's a good idea actually and I've done some research on it. You need to estimate following parameters in order to detect drowsiness:

1. Blink frequency
2. Blink duration
3. Percentage of eye closure
4. Velocity of eyelid movements (opening and closing)

Obviously you don't need to use all of them, but it's better not to rely on any single value and use bunch of these to increase accuracy. I'm not sure if it's possible to derive these using current version, so would request TheEyeTribe officials to throw some light on this.

suryotrisongko@gmail.com wrote:I was wondering, is it possible to develop a drowsiness and distraction monitoring system ( http://www.mynewsdesk.com/us/tobii_tech ... ems-755160 ) using theeyetribe eye tracking device?

Can anyone from TheEyeTribe team please throw some light on this question?

We currently do not provide blink states through the API, although it's something we have in our roadmap (currently no ETA for this). It is however possible to detect blinks by using the information provided through the API, namely eye detection and face detection states.

For instance, during blinks the eyes are not detected for a few samples, but during those samples the face is detected. Once the blink is finished, the pupils are detected again more or less in the same location as before the blink. You could write a piece of software that keeps track of these state changes and assigns a "blink event" when such sequence takes place. The downside is that the blink event is assigned a posteriori, e.g. once it has already happened. Depending on your application, this might be good enough.