Hello, I am developing a grasping algorithm using your dataset. I would like to ask how are the training and test sets in your data set divided?