Can I use a pre-trained YOLO neural network to recognize objects in images to classify all the frames in a video and then add those classified images to the original dataset to retrain the same network?
I'm wondering if it could be a valid method to expand my dataset or if it's a case of "'the dog chasing its own tail".