The videos in our dataset are named as:
RGB_actionname_subjectname_groupnum.avi or D_actionname_subjectname_groupnum.avi

In order to provide a standard experimental setting, 
we formally defined two evaluation protocols, including cross-group (CG) and cross-subject (CS) settings. 

In the CG setting, 
considering each subject performed each action for 3 times, 
we divided the videos samples into 3 groups(according to the names of the videos), 
where a group was for training and the rest for testing.

In the CS setting, 
we separated the 8 subjects into 4 splits and used samples from 3 splits for training and the other for testing. 
splits1: cgy, cwx
splits2: fjw, gs
splits3: lhh, lzy
splits4: tys, wyy