Microexpression Pain Detector

A webcam app that tries to detect pain from facial expressions using computer vision. I built this to learn about FACS-based pain detection and real-time video processing.

---

What it does

Uses your webcam to track facial muscle movements and estimate whether someone might be in pain. Based on the Facial Action Coding System (FACS) — looking for specific action units like brow furrowing, eye squinting, and nose wrinkling that consistently appear in pain expressions across clinical research.

Not a medical tool. Student project.

---

Setup

pip install -r requirements.txt
python -m streamlit run app.py --server.port 8502

Go to http://localhost:8502

---

How to use

1. Click "Start Webcam"
2. Let it detect your face
3. Click "Set Baseline" with a neutral expression
4. Make different expressions to see the detection respond

---

Project structure

src/
  detection/      face detection and feature extraction
  models/         CNN model (optional, rule-based works for testing)
  visualization/  drawing overlays on the video feed
app.py            main Streamlit app
train_model.py    train the CNN yourself if you want

---

Requirements

• Python 3.8+
• Webcam
• Packages in requirements.txt

---

Limitations

Lighting affects accuracy significantly. Works differently across people. Not validated on clinical populations. The rule-based approach is good enough for exploring how the system responds — the CNN model is optional and needs its own training data to be useful.

---

Citation

Youssof Sallam. Microexpression Pain Detector (2025). GitHub. https://github.com/youssof20/microexpression-pain-detector