🦷 DentaScanAI – AI-Powered Dental Health Analysis

---

🚀 Overview

DentaScanAI is an AI-powered web application built to assist in dental health analysis through image-based diagnosis.
It uses Deep Learning (YOLOv8-L) to detect and classify dental issues from X-ray images and integrates real-time doctor consultation, AI-generated health reports, and clinic recommendations — all within one platform.

This project was developed during MEDHA-2025 (IIT Bombay Edition) Hackathon hosted by GL Bajaj Institute of Technology & Management, as part of a 36-hour hackathon sprint.

---

🧩 Project Features

• 🧠 AI Model (YOLOv8-L) trained on 2400+ dental X-ray images across 9 categories.
• ⚙️ 80-20 Train-Test Split with 48 epochs of training.
• 📈 Achieved ~75–80% accuracy during evaluation.
• 🩺 AI-Powered Diagnosis: Detects and classifies dental conditions automatically.
• 🧾 AI-Generated Reports: Summarizes detected issues and recommended next steps.
• 👩‍⚕️ Doctor Connect: Real-time chat and video consultations with dentists.
• 🏥 Nearby Clinics: Suggests local dental clinics for quick appointments.
• 💬 Home Remedies: Offers temporary relief suggestions based on detected issue.

---

⚙️ Tech Stack

Component	Technology
Model	YOLOv8-L (Ultralytics)
Language	Python
Libraries	OpenCV, NumPy, Pandas, Torch, Ultralytics
Frontend	HTML, CSS, JavaScript, React (optional)
Backend	Flask / FastAPI
Database	MongoDB / Firebase
Deployment	Local Server (Hackathon Build)
GPU Used	NVIDIA GeForce RTX 3050 (6 GB)

---

📊 Model Performance

Metric	Result
Training Images	2400
Epochs	48
Accuracy	~75–80%
Dataset Split	80% Train / 20% Test

---

🔄 Workflow

graph TD
A[Dataset Collection & Preprocessing] --> B[Annotation & Labeling]
B --> C[YOLOv8-L Model Training]
C --> D[Evaluation & Accuracy Testing]
D --> E[Integration with Backend API]
E --> F[Frontend Web Interface]
F --> G[Real-time Inference & User Interaction]

Loading