SkinSense AI

AI-powered skin analysis assistant that combines Computer Vision, Deep Learning, and Generative AI to analyze facial skin conditions and provide personalized skincare recommendations.

---

Overview

SkinSense AI allows users to:

• Upload a facial image
• Capture a photo using a webcam
• Detect faces automatically
• Predict skin conditions using a Deep Learning model
• Generate AI-powered skincare recommendations
• Download a PDF skin analysis report

---

Features

Computer Vision

• Face Detection using MediaPipe
• Automatic Face Cropping
• Image Preprocessing
• Webcam Support
• Facial Image Analysis

Deep Learning

• MobileNetV2 Transfer Learning Model
• Skin Condition Classification
• Confidence Score Generation
• Top 3 Predictions
• Severity Estimation

AI Recommendations

• Powered by Google Gemini
• Condition Explanation
• Possible Causes
• Recommended Skin Care
• Ingredients to Look For
• Things to Avoid
• When to Visit a Dermatologist

PDF Reports

• Downloadable Skin Analysis Report
• Prediction Summary
• Confidence Score
• Severity Level
• AI Recommendations

User Interface

• Streamlit Web Application
• Image Upload Support
• Webcam Capture Support
• Interactive Results Display

---

Supported Skin Conditions

The model can classify the following skin conditions:

• Blackheads
• Dark Spots
• Inflammatory Acne
• Pigmentation
• Pores
• Redness
• Whiteheads
• Wrinkles

---

AI Model

SkinSense AI uses a MobileNetV2-based Transfer Learning Model trained for skin condition classification.

Model Architecture

Input Image (224×224×3)
        ↓
MobileNetV2 (Pretrained on ImageNet)
        ↓
GlobalAveragePooling2D
        ↓
Dropout
        ↓
Dense (128)
        ↓
Dropout
        ↓
Dense (8 Classes)

Model Details

• Base Model: MobileNetV2
• Framework: TensorFlow / Keras
• Input Size: 224 × 224
• Number of Classes: 8
• Transfer Learning: Yes

---

Tech Stack

Frontend

• Streamlit

Computer Vision

• OpenCV
• MediaPipe
• Pillow

Deep Learning

• TensorFlow
• Keras
• MobileNetV2

AI Integration

• Google Gemini API

Data Processing

• NumPy

Report Generation

• ReportLab

---

Project Structure

SkinSense-AI/
│
├── app.py
│
├── ai/
│   └── recommendation.py
│
├── models/
│   ├── predictor.py
│   └── model.keras
│
├── reports/
│   └── report_generator.py
│
├── utils/
│   └── face_detection.py
│
├── assets/
│
├── requirements.txt
├── README.md
└── .env

---

Installation

1. Clone Repository

git clone https://github.com/Diksha6524/SkinSense-AI.git

2. Navigate to Project Folder

cd SkinSense-AI

3. Create Virtual Environment

python -m venv venv

4. Activate Virtual Environment

Windows

venv\Scripts\activate

5. Install Dependencies

pip install -r requirements.txt

---

Gemini API Setup

Create a .env file in the root directory.

GEMINI_API_KEY=YOUR_API_KEY

Get your API key from Google AI Studio.

---

Run the Application

streamlit run app.py

---

Application Workflow

Upload Image / Webcam Capture
              ↓
        Face Detection
              ↓
         Face Cropping
              ↓
      Image Preprocessing
              ↓
     MobileNetV2 Prediction
              ↓
     Skin Condition Analysis
              ↓
     Gemini AI Recommendation
              ↓
      PDF Report Generation
              ↓
        Result Display

---

Screenshots

Home Page

Prediction Results

PDF Report

---

Future Enhancements

• Progress Tracking Dashboard
• Before vs After Comparison
• User Authentication
• Skin Health History
• Cloud Deployment
• Mobile Application
• Personalized Skincare Planner
• Product Recommendation Engine
• Dermatologist Consultation Integration

---

Learning Outcomes

Through this project, I learned:

• Computer Vision using OpenCV
• Face Detection using MediaPipe
• Deep Learning with TensorFlow
• Transfer Learning using MobileNetV2
• Gemini API Integration
• Streamlit Web Development
• PDF Report Generation
• End-to-End AI Project Development

---

Disclaimer

This project is intended for educational purposes only.

The predictions and skincare recommendations generated by SkinSense AI should not be considered medical advice and do not replace consultation with a qualified dermatologist.

---

Author

Diksha

Computer Engineering Student

---

Support

If you found this project useful, please consider giving it a ⭐ on GitHub.