I am currently working on face detection using alt and jones.
I think I'm making some stupid assumptions that end with the 100% FALSE negative rate and 0% true positive rate classifiers.
Here's how I created my training functions: for all four types of functions, I generated different sizes, but equal width and height for the black and white area.
Could this be the reason for the unsuccessful learning outcome?
Moreso, my threshold for each function was arbitrarily selected within a range of values.
Am I really wrong? I would be so happy if someone could help. I spend so much time on this and the exam is nearby.
source share