SkinAI – AI-Powered Skin & Acne Analyzer

Live Demo: https://acnemodel.streamlit.app

---

Overview

SkinAI is a real-time facial skin analysis tool that detects:

• Acne (pimples, blackheads, whiteheads)
• Skin tone & undertone
• Likely ancestral origin (continent genes)
• Skin type (Oily / Dry / Combo)
• Smoothness score (0–100)

It uses computer vision + color science to give instant dermatology-level insights from a selfie.

---

Features

Feature	Description
Face Detection	Auto-detects face using OpenCV Haar Cascade
Acne Detection	Finds red, white, blackheads with numbered red circles
Location Mapping	Labels acne as: Forehead, Left/Right Cheek, Chin, Jaw
Skin Tone Analysis	Classifies: Very Fair → Deep Brown
Undertone	Warm (Golden) vs Cool (Pink)
Genetic Origin (Approx)	Europe, Mediterranean, South Asia, Africa, East Asia
Skin Type	Oily / Combination / Dry based on brightness variance
Smoothness Score	Texture analysis (0–100)
Zoomable Acne Crops	Click each pimple to see close-up
Web App (Streamlit)	Upload selfie → Instant report

---

Tech Stack

Layer	Technology	Version / Notes
Frontend / UI	Streamlit	1.30+ – Interactive web app
Backend / Logic	Python	3.11
Computer Vision	OpenCV	opencv-python-headless (cloud-safe)
Image Processing	NumPy, PIL (Pillow)	Core array & image handling
Color Analysis	HSV & Lab color spaces	For tone, acne, texture
Face Detection	Haar Cascade (haarcascade_frontalface_default.xml)	Auto-downloaded
Deployment	Streamlit Community Cloud	Free, auto-deploy from GitHub
Version Control	Git + GitHub	git
Package Management	requirements.txt	Cloud dependency install
Local Dev	Conda (Miniconda)	skinai environment

---

Project Structure

SkinAI/
│
├── app.py                  # Main Streamlit app (all logic)
├── requirements.txt        # Cloud dependencies
├── haarcascade_frontalface_default.xml  # Auto-downloaded
├── README.md               # This file
└── .git/                   # Git repo

---

How It Works (Technical Flow)

1. Upload Selfie → PIL → NumPy array → OpenCV (BGR)
2. Face Detection → Haar Cascade → Crop face ROI
3. Skin Mask → HSV range → Remove eyes/lips (anti-mask)
4. Acne Detection:
   • Red inflamed: High saturation + value
   • Dark scars: Low value
   • Shape filter: Circularity > 0.5
5. Color Analysis → Mean HSV in skin mask
6. Texture → Variance in grayscale → Smoothness score
7. Output → Streamlit UI with images, metrics, zoomable crops

---

Local Setup (Mac / Linux / Windows)

# 1. Clone repo
git clone https://github.com/yourusername/SkinAI.git
cd SkinAI

# 2. Create conda env
conda create -n skinai python=3.11 -y
conda activate skinai

# 3. Install deps
pip install -r requirements.txt

# 4. Run locally
streamlit run app.py

Open http://localhost:8501 → Upload selfie → See results!

---

Deploy to Streamlit Cloud (Free)

1. Push to GitHub
2. Go to share.streamlit.io
3. New App → Link your repo → app.py → Deploy!
4. Done in 60 seconds!

---

Screenshots

Screenshot	Description
	Upload + Face + Acne Circles
	Color, Genes, Type, Smoothness

---

Limitations & Future Ideas

Current Limit	Future Upgrade
Lighting-sensitive	Add auto-light correction
Single face only	Multi-face support
No age/gender	Add CNN models (MediaPipe)
No history	Add user login + save reports
No PDF export	Add "Download Report" button

---

Author

Shubham

• GitHub:

"Turning selfies into skin science — one pimple at a time."

---

Star This Project!

If you like it, give it a star on GitHub!
Help others find this AI dermatology tool.

---

SkinAI — Because your skin deserves AI.