Beetle Tracker is a small lightweight pair of scripts used to tag beetle videos and then track them using idtrackerai. Analysis is completed after tracking and various files are generated
Beetle tracker runs in two phases. First a video must be tagged manually by a user. This generates a file which is used in the track phase. The tag phase can be run on virtually any machine, but the track phase should be run on a machine with a CUDA runtime.
- Open the ui with
python tag/GUI_preprocessing.py - Click on the buttons until each requirement is complete. Follow onscreen directions or hints in the console
- If tracking on a different computer than tagging on, move the video and generated
.jsonfile to the tracking machine
- Run the tracker tool
python track/postprocessing.py <filename.json>
prereq: python 3 installed (Tested with 3.10)
$ python --version
Python 3.10.12git clone git@github.com:jsngalloway/beetle-tracker.git
cd beetle-tracker/tag
# Create and activate a python virtual environment for tagging
python3 -m venv ./venv
source ./venv/bin/activate
pip install -r requirements.txt
# Test your installation
python GUI_preprocessing.py
# End your session
deactivateThe beetle tracker tagging UI looks like this:
Before installation ensure the CUDA runtime is configured. (nvidia docs) Verify installation with nvidia-smi it should output something like:
Wed Jan 1 08:46:51 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 560.35.05 Driver Version: 560.35.05 CUDA Version: 12.6 |
|-----------------------------------------+------------------------+----------------------+
...
Install the requirements for beetle tracking:
git clone git@github.com:jsngalloway/beetle-tracker.git
cd beetle-tracker/track
# Create and activate a python virtual environment for tracking
python3 -m venv ./venv
source ./venv/bin/activate
pip install -r requirements.txt