I see two methods how to train model:
- training based on one picture with my true predict of decision
- training based on one big picture with set of pics inside and keep training like sets with my true predict of decision
Predict input may be as one picture or sets
If you have suggestions, will be nice to hear your best practice for how to properly give training input how to properly give predict input
Now I keep training my model with inputs as sets and big separated images. I receive answers based on one picture or sets as predict input.