🧠 DermaScan – AI-Based Acne Severity Detection App

DermaScan is a mobile-ready deep learning application that classifies acne severity into Mild, Moderate, and Severe using a lightweight CNN model (MobileNetV2). The system is optimized for on-device inference using TensorFlow Lite, ensuring privacy and real-time performance.

---

🚀 Features

• 📷 Capture or upload skin images
• 🤖 AI-based acne severity classification
• 📊 Outputs severity + confidence score
• 📱 Runs offline using TFLite (no internet required)
• 🔒 Privacy-first (no data leaves device)
• 🌍 Optimized for Indian skin tones

---

🧠 Model Details

• Architecture: MobileNetV2 (Transfer Learning)
• Input Size: 224 × 224 RGB images
• Classes:
  • Mild
  • Moderate
  • Severe
• Optimization: TFLite Quantization
• Dataset: DermaCon-IN (Indian dataset)

---

🔁 System Pipeline

Steps:

1. Image Capture / Upload
2. Preprocessing (Resize + Normalize)
3. Model Inference (MobileNetV2)
4. Severity Prediction + Confidence

---

🏗️ System Architecture

The system processes user input locally and returns acne severity without requiring cloud interaction.

---

📊 Data Flow Diagrams

Level 0 DFD

Level 1 DFD

---

🔄 Workflow (State Diagram)

---

🔧 Modules

• User Interaction Module → Capture/upload images
• Image Processing Module → Resize, normalize
• Classification Module → MobileNetV2 prediction
• On-Device Inference Module → TFLite execution

---

🧪 Testing

• ✅ Unit Testing (Preprocessing, Model, TFLite)
• ✅ Integration Testing (End-to-End pipeline)
• ⚡ Real-time performance tested (<2 seconds)

---

📱 Screenshots

🏠 Home	📤 Upload	🖼️ Selected	⚙️ Analyzing	📊 Results

🛠️ Tech Stack

• Python
• TensorFlow / Keras
• TensorFlow Lite
• OpenCV / PIL
• Android Studio
• Google Colab

---

👥 Team

• Abhirami Sajith
• Afia Nasumudeen
• Akshara Cheruvatheri
• Alex Mathai Mathews