Thank you for the outstanding work. I would like to ask if the model for this task can be applied to regular 2D images?