abhishaike / asl_translation Goto Github PK

Look into K-Means, or other clustering algos, to find good color ranges to segment out a hand.

Function must create dynamic (as in, changeable) and SQUARE windows, given an image. You will not know the image size beforehand, so account for that in your function. Ideally, try to use a generator for this.

Investigate what type of classifier we want to end up using. Right now, I'm thinking SVM, but there may be faster options out there. This investigation should be done for both hand/not-hand (binary) and ASL translation (multi class). Remember, we're using HoG features.