KTH, Weizmann, and IXMAS datasets

The KTH, Weizmann, and IXMAS datasets are used for action recognition. The KTH dataset contains 600 videos that are made up of 25 subjects performing 6 actions in 4 scenarios. The Weizmann dataset contains 90 videos of 9 subjects performing 10 actions. The IXMAS action recognition dataset is made up of 10 subjects, 11 actions and 1148 sequences.