I have a face detection algorithm that works great and fast (IOS 'default CID detector) when I use it Real Human Face Now I am finding a test-image that I had placed in a room elsewhere on the screen, so I have complete control of image-to-detection, only one thing I should be sure about On size
I tested it with a simple smiley depicted on the HTML5 canvas (like: Here is what is identified as a face, but not as fast and sometimes as a real image (like: .
As you can expect, it is actually difficult to do Google, is it really clear facial features There is a goto-schematic picture that people use?
EDIT: If someone knows a good algorithm for reverse-engineer, the correct image for an existing black-box face-recognition , This will also solve the problem;)
No comments:
Post a Comment