Train models with images and text
Create images from various types of annotations
Describe images using multiple models