Do you have to do image sequence tagging, is to split the video image sequence, and then according to the category label input to the model, made of python