Here, I train a neural network on images that I recorded of my hand, to identify how many fingers I am holding up - resulting in an almost 100% accuracy. This is then used for a localisation script, which finds an image of my hand in a larger image. This can be generalised to multiple pictures of hands in one image.
