1. Datasets location:
	Leg Band: ./data/LegDay/
	Robot Skin: ./data/RoboSkin/
	Table Cloth: ./data/table_v2/
	note: ./data/Smart-Table/ is legacy and raw data. Because the plates need to be segmented in the orignial data

2. TPM Feature set:

frame descriptors: frame2des.m
from frame descriptors to temporal features: fempfeatrues.m
from data stream to temporal features: stream2tempfeats.m
key frames: stream2kf.m
from key frames to spatial features: frame2moments.m
from data stream to spatial features: stream2spacfeats.m

3. brancedNCA.m, KFoldNCA.m are the NCA division methods

4.  ./ltfat/ is the library needed for fast wavelet transform

5. for the paper:
script_dining_feat_eval_p4.m and script_dining_feat_eval_p4.m for the table cloth dataset,
script_legday_feat_eval_p1.m and script_legday_feat_eval_p2.m for the leg band dataset,
script_roboskin_feat_eval_p1.m and script_roboskin_feat_eval_p1.m for the robot skin dataset
are the files used to generate the results in the paper.