Najran University Graduation Project
AI-Powered Skin Analyzer
& Makeup Recommender
A mobile app that analyzes skin type, undertone, and concerns from a selfie — then recommends skincare and makeup products, with an admin dashboard for catalog & logs.
⚡ ≤ 5s Analysis
🔒 Privacy-First
🎯 Personalized
Skin Type
Oily / Dry / Combo
Undertone
Warm / Cool / Neutral
Modules
AI + AR + Admin
🧠 TensorFlow + MediaPipe
💄 Shade Matching
📊 Logs & Metrics
Key Features
Designed as an all-in-one AI beauty platform.
🧬
Skin Type & Concerns
Detect oily/dry/combination/sensitive and issues like acne, pigmentation, wrinkles, redness.
🎨
Undertone Detection
Warm/cool/neutral classification using color-space analysis + AI, improving foundation matching.
🧠
Personalized Recommendations
Rule-based + scoring engine that maps results to skincare ingredients and makeup shades.
🛠️
Admin Dashboard
Manage brands, products, categories, shades, and review AI logs with secure access.
🔐
Privacy Controls
No unnecessary storage; secure processing; clear disclaimers (not a medical diagnosis).
🌍
Bilingual UI
Arabic/English with full RTL support + dark/light modes.
How It Works
Simple flow from selfie to recommendation.
1
Capture / Upload
User takes a selfie under good lighting or uploads a clear image.
2
AI Analysis
Preprocessing + CNN models + undertone estimation generate scores and confidence.
3
Recommend + Try-On
Return products & shades; optionally preview makeup with AR try-on.
✨ What makes it “smart”?
• Face landmarks isolate cheek/jaw regions for undertone
• CNN detects multiple concerns simultaneously
• Match scores rank products & shades with clear reasons
• CNN detects multiple concerns simultaneously
• Match scores rank products & shades with clear reasons
🧪 Prototype Focus
Start with core modules: analysis + product catalog + recommendations.
Then expand to reviews, analytics, and advanced AR optimization.
System Architecture
Mobile + Backend + Database + Admin dashboard — modular & scalable.
📱
Mobile App (Flutter)
Cross-platform UI, camera capture, image validation, results screens, recommendations, language & theme.
🧩
Backend (Django REST)
Secure APIs (/api/analyze, /api/products), authentication, business logic, logs, and integrations.
🧠
AI Engine
MediaPipe Face Mesh + OpenCV preprocessing + TensorFlow CNN models (MobileNet/EfficientNet).
🗄️
Database (MySQL)
Users, analysis history, products, shades, match scores, reviews, logs, analytics tables.
🛡️
Security & Privacy
• HTTPS/TLS for all traffic
• Password hashing (PBKDF2)
• Optional private analysis & retention controls
• Password hashing (PBKDF2)
• Optional private analysis & retention controls
⚡
Performance
• Model optimization (quantization / TFLite later)
• Cache repeated computations
• AR target: ≥ 25 FPS (mid-range devices)
• Cache repeated computations
• AR target: ≥ 25 FPS (mid-range devices)
Screens Preview
Placeholder visuals (replace with your real UI screenshots).
Analysis Result
Recommendations & Shades
Admin Dashboard & Logs
Ready to explore the dashboard?
Open /admin to manage products, logs, and analytics for the AI beauty platform.
Open Admin
FAQ
Common questions about the project.
Is this a medical diagnosis app?
No. It provides cosmetic guidance and skin insights; it does not replace dermatologists.
What affects accuracy?
Lighting, blur, face angle, and camera quality. The app validates image quality before analysis.
Can we add more brands/products later?
Yes. The admin dashboard is designed to expand the catalog and manage categories/shades easily.