Getting started¶
There are three ways into this project, and they do not depend on each other. Pick the one that matches what you have in front of you.
-
Only a laptop
Install the toolkit and run the live dashboard against your webcam. The detection pipeline, the risk logic and the tuning UI are all real; only the board is missing.
-
A board on the desk
Build, flash and monitor the firmware with
./plxy.sh dev, then join the board's access point to see what it sees. -
A box of parts
Start at the hardware tutorial: four components, seven wires, and a test after every stage.
What you need¶
| For | You need |
|---|---|
| Desktop toolkit, live dashboard | Python 3.10+ and a webcam |
| Training and export | the above, plus a dataset (see Datasets) |
.espdl quantization |
the above, plus esp-ppq |
| Firmware | ESP-IDF 5.x on Windows, macOS or Linux, and the board |
| Documentation | Python 3.10+ only — see the documentation pipeline |
The 60-second version¶
git clone https://github.com/SengPhirum/PLXY_DrowsyGuard.git
cd PLXY_DrowsyGuard
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\Activate.ps1
pip install -e ".[live]"
drowsyguard doctor # what is installed, what is missing
drowsyguard fetch-models # YuNet detector + eye-state model
python -m drowsyguard.cli live # then open http://127.0.0.1:8000
No checkpoint is needed: eye mode is the default and uses the downloaded eye-state model.