CUBOIDS/RECOGNITION See also Demo recognition_demo - Describes all steps of behavior recognition; example for facial expressions. datadir - Get location of data; alter file depending on location of dataset. Feature detection and description. featuresLG - Feature detection and description applied to .mat behavior data. featuresLGconv - Convert features to DATASETS format so output is same as after featuresSM. featuresLGdesc - Applies descriptor to every cuboid of every clip of every set. featuresLGdetect - Detects features for each set of cuboids using stfeatures. featuresLGpca - Dimensionality reduction for cuboids descriptors. featuresSM - Feature detection and description applied to DATASETS behavior data. featuresSMdesc - Applies descriptor to every cuboid of every clip of every set. featuresSMdetect - Detects features for each set of cuboids using stfeatures. featuresSMpca - Dimensionality reduction for cuboids descriptors. Conversion between various data formats. conv_clips2datasets - Converts between representations of behavior (mat -> DATASETS). conv_clips2movies - Converts between representations of behavior (avi -> mat). conv_datasets2clips - Converts between representations of behavior (DATASETS -> mat). conv_movies2clips - Converts between representations of behavior (mat -> avi). conv_movies2divx - Converts between representations of behavior (avi -> compressed avi). Classification. recog_clipdesc - Creates a clip descriptor that is simply a histogram of cuboids present. recog_clipsdesc - Create descriptor of every clip. recog_cluster - Clusters all cuboids in DATASETS (based on their descriptions). recog_test - Test the performance of behavior recognition using the cuboid representation. recog_test_nfold - Test the performance of behavior recognition using cross validation.