face-mask-detection | Simple project to detect if person | Machine Learning library

I think you got some misunderstanding about how indexing and slicing in numpy works:

For your first line y:y + h basically means slice from index y to index y + h. Same goes for the second dimension, there you take the array slice from x to x + w.

So basically x and y correspond to your xmin and ymin and you slice to xmax and ymax which is the window minimums plus the corresponding height and width; that's why you have y + h and x + w in the second part of both dimensions.

As for frame[y:y, x:x] - this makes no sense at least in python, so you might have mixed up something there. If you slice from the index to the same index, your resulting array will always be empty: