but when i try to run my detection in the photos that i put it in the training process it doesn't recognize them more than a few times no matter how many times i edit the object . also the problem is that when i try to fix the rotation for example of the detection box it messes up all the other cases that worked.