deep convolution inverse graphics network (dc-ign)

A model used to understand the key factors of object structure in images. It's most often used in computer vision tasks.