Ouch, not quite18.9041and11.7159we were after. But hey, we forgot to scale the features! Let's calculate the L2 norm (i.e. Euclidean norm / distance) and divide the features first so they become unit vectors: >>> # normalized features >>> image_feature...