Idea is to create object bounding boxes and then use that to determine where the camera needs to move. This doesn't need to be as complex for a first pass POC. Once the base model is created, focus on centering and finding the best angles can be implemented.
Looking at:
https://www.pyimagesearch.com/2017/09/11/object-detection-with-deep-learning-and-opencv/
https://www.pyimagesearch.com/2018/11/12/yolo-object-detection-with-opencv/
Idea is to create object bounding boxes and then use that to determine where the camera needs to move. This doesn't need to be as complex for a first pass POC. Once the base model is created, focus on centering and finding the best angles can be implemented.