How to add image as observation when using Learning Agents

Yes, that is what i intended to do. I have used airsim plugin from microsoft and it work with image input, but it’s limited to one agent at a time. That’s why i want to find a new method.