load_raw_labels

behavenet.data.preprocess.load_raw_labels(file_path, pose_algo, likelihood_thresh=0.9)

Load labels and build masks from a variety of standardized source files.

This function currently supports the loading of csv and h5 files output by DeepLabCut (DLC) and Deep Graph Pose (DGP).

Parameters:

• file_path (str) – absolute file path of label file

• pose_algo (str) – ‘dlc’ | ‘dgp’

• likelihood_thresh (float) – likelihood threshold used to define masks; any labels/timepoints with a likelihood below this value will be set to NaN and the corresponding masks file with have a 0

Returns:

• (array-like): labels, all x-values first, then all y-values

• (array-like): masks; 1s correspond to good values, 0s correspond to bad values

Return type:

tuple