Table 1 Sample data quantification.

From: Surface and underwater human pose recognition based on temporal 3D point cloud deep learning

Pose

Number of decomposition actions

Number of point clouds per action

Total number of point clouds

Breaststroke

8

50

400

Butterfly stroke

8

50

400

Backstroke

8

50

400

Freestyle

8

50

400

Tread water

8

50

400

Drown

8

50

400