BehaveTrack uses user-specific B-SOiD/A-SOiD classifier to predict behavior based on pose. Generate ethogram and summary statistics for animal recordings. Intuitive interface for tagging and grouping files. Enhance animal behavior studies with accurate and reliable predictions.
create a conda environment
conda create -n behavetrack python==3.9 -y
conda activate behavetrackclone github repo and change directory into repo
git clone https://github.com/runninghsus/BehaveTrack.git
cd BehaveTrackinstall dependencies
pip install -r requirements.txtRun the streamlit app
streamlit run behavetrack.pyQuality of life improvements:
- Drag and drop B-SOiD trained classifier
- Drag and drop A-SOiD trained classifier
- Using example video to annotate user-definition
- Organize files into 2+ conditions for comparisons, including the following:
Reactive post-hoc analyses:
- ethogram
- behavior location
- behavior ratio
- behavioral frequency
- behavior duration
- behavior transition
Quality of life improvements:
- annotation retrieval from A-SOiD
Reactive post-hoc analyses:
- behavioral pose speed
Glen de Vries Presidential Fellowship for Biological Sciences at Carnegie Mellon University.
This is currently developed by Alex Hsu to support research using B-SOiD.
BehaveTrack is released under a BSD 3-Clause "New" or "Revised" License and is intended for research/academic use only.