SBAUtil 0.1.0

Semi-automatic Pipeline for Markerless Pose Estimation and Behavior classification of mouse.

SBA

Untility functions for interaction classifcation. The module provides functions for pre-SIMBA and post-SIMBA processing.

Installation

pip install SBAUtil

Tutorial

import SBA

Initalize the configuration file:

SBA.initConfig('./',projectName = "newProject", experimenter = "Yiyang")

dynamic cut

SIMBA gives error if the h5file input from DLC contains a portion where there is only one mouse. SBA.batch_dynamic_cropVideo will

1. dynamically crop out this portion based on the given h5file from DLC.
2. modify the original DLC h5 file to match the extracted video.
3. The new videos and new h5files can then be used for SIMBA analysis

SBA.batch_dynamic_cropVideo(config_path,videos,h5files,videoType="avi")

videos: The path to directory of the videos.

h5files: The path to the directory of the DLC h5 outputs.

Post SIMBA processing

finalizeSimBaOutput does the following:

1. Simplify the final output from SIMBA.
2. Modify the frame number field so that it matches the original video.

SBA.finalizeSimBaOutput(config, simbaFiles, classifier ,destDir = None)

config: path to the SBA configuration file.

simbaFiles: path to the SIMBA output directory. It's by default in project_folder/csv/machine_results.

classifier: list of strings containing the classifier names. For example, ["isInteraction"] or ["Unilateral", "Reciprocal", "Passive"]. Make sure the spelling matches the names defined in the SIMBA model.