Software to adjust brightness and contrast of behavior task videos recorded in low light. A neural network (DeepLabCut) is used to identify body parts of each rat (Head/Neck/Tail base) to be used in further analysis. Matlab code filters the position data for each body part and calcualtes the angle of the rat's body in each frame. Body angle is used to determine periods of increase looking behavior that will be integrated into other analysis pipelines.
Rotation_Sessions.xlsx
Spreadsheet with rats and sessions to analyze.
An anaconda environment to use with DeepLabCut for novel video analysis is included in
Rodent-Turns\DLC\Anaconda Environment\environment.yml
Nvidia CuDnn libraries will need to be added to the enviornment's directories. The CNN model is stored in
Rodent-Turns\DLC\ICR Behavior-Gia-2020-11-23\dlc-models\iteration-1\ICR BehaviorNov23-trainset85shuffle1\train\snapshot-700000.7z.001
This file should be unzipped with 7zip and the
Rodent-Turns\DLC\ICR Behavior-Gia-2020-11-23\config.yml
path variable should be updated before use.
sessionIterator.py
Is the script to iterate through directories of interest and run the processVideo.py
function to adjust brightness and contrast of video and analyze with DeepLabCut. CheckLabels.py
generates labeled videos for VT1 videos.
sessionIterator.m
Is the script to iterate through directories of interest and run the TurnDetection.m
function to filter position coordinates, calculate body angle, and detect scanning events in the video. Can also choose to call CreateTurnsVideo.m
to create a labeled video to visualize detected scanning events for a given session.