From Machine Perception to Machine Learning: generating models of visual objects