🦷 Teeth Classification using CNN

AI-Powered Dental Image Classification System

---

Live Demo

🔗 Streamlit App:
https://teeth-classification-cnn.streamlit.app/

---

Project Overview

This project presents an end-to-end AI-powered dental image classification system built using a Convolutional Neural Network (CNN).

The model classifies dental images into 7 distinct oral disease categories, forming a baseline intelligent diagnostic support tool for dental healthcare applications.

This project represents the first phase of a larger AI-driven medical initiative focused on improving diagnostic accuracy in dental imaging.

---

Objectives

• Preprocess and normalize dental images for stable CNN training
• Apply data augmentation to enhance generalization
• Analyze dataset balance using visualization techniques
• Build a CNN model from scratch using TensorFlow
• Establish a strong baseline performance
• Deploy the trained model using Streamlit Community Cloud

---

Disease Classes

The model classifies images into the following 7 categories:

• CaS
• CoS
• Gum
• MC
• OC
• OLP
• OT

---

Model Architecture

A CNN model was built from scratch with:

• Multiple Conv2D layers for feature extraction
• MaxPooling2D for spatial downsampling
• GlobalAveragePooling2D to reduce parameters
• Fully connected Dense layers for classification
• Dropout layers to reduce overfitting

Architecture Diagram

---

Dataset & Preprocessing

All images were:

• Resized to 256 × 256
• Normalized to pixel range [0, 1]
• Augmented using:
  • Rotation
  • Horizontal flipping
  • Zooming

These techniques improve robustness and reduce overfitting.

---

Training Performance

The model achieved:

• Strong training accuracy
• Stable validation accuracy
• Smooth loss convergence
• No significant overfitting

Training vs Validation Accuracy

---

Deployment

The trained model was deployed using:

• Streamlit
• TensorFlow 2.20
• Streamlit Community Cloud

Users can:

• Upload a dental image
• Receive predicted disease class
• View model confidence score

---

Project Structure

Project Structure

teeth-classification-cnn/
│
├── app.py # Streamlit application
├── requirements.txt # Dependencies
│
├── images/ # Visual assets
│ ├── model.png
│ ├── output.png
│ └── training_validation_accuracy.png
│
├── model/
│ └── best_model.h5 # Trained CNN model
│
├── dataset/ # Processed dataset
├── notebook/ # Jupyter notebooks
└── pdf_task/ # Project documentation

---

Technologies Used

• Python
• TensorFlow / Keras
• NumPy
• Matplotlib
• Streamlit
• Git & GitHub

---

Author

Armia Gamal
AI & Computer Vision Enthusiast

🔗 GitHub: https://github.com/Armia-Gamal
🔗 LinkedIn: https://www.linkedin.com/in/armia-gamal/

---

Final Note

This project demonstrates the practical integration of:

• Deep Learning
• Medical Imaging
• Model Deployment

into a real-world AI healthcare application.