Closed qxcv closed 8 years ago
Training poses should be jittered be a random amount in [-step, step] along each axis to ensure that biposelets which are similar can be distinguished.
[-step, step]
Training poses should be jittered be a random amount in
[-step, step]
along each axis to ensure that biposelets which are similar can be distinguished.