I was playing with your demo page and was curious, how do you handle gender classification? Follow-up would be what are you results in broad ethnicity recognition (based on what I've seen, I can tell you at least tried it). Any attempts into age?
We have trained our system so that it knows typical/common female and male facial features.
Our machine learning algorithm will generate a score from 0.0 to1.0 representing gender.
We are still tuning our age and emotion recognition. We will release (most likely in just a couple of weeks) it when we feel it's good and reliable enough for public use.
