Defect Detection Using Autoencoders and FCM

Overview

This project implements defect detection using Autoencoders and Fuzzy C-Means (FCM) clustering instead of traditional K-Means clustering. The model is designed to classify images into defected and non-defected categories.

Features

• Uses Autoencoders for feature extraction.
• Employs FCM clustering for classification.
• Supports image-based defect detection.
• Flask web interface for uploading and analyzing images.

Repository Structure

📂 project_root/
 ├── 📂 templates/            # HTML templates for UI
 ├── 📄 main.py               # Main application logic
 ├── 📄 README.md             # Documentation
 ├── 📦 Defected.zip          # Dataset: Defected images
 ├── 📦 Non_defected.zip      # Dataset: Non-defected images

Requirements

Install dependencies using:

pip install -r requirements.txt

Usage

1. Extract datasets from Defected.zip and Non_defected.zip.
2. Run the Flask app:

   python main.py

3. Access the Web Interface: Open http://127.0.0.1:5000/ in your browser.

Algorithm: Autoencoders + FCM

Instead of K-Means, this project uses Fuzzy C-Means (FCM) for clustering defected vs. non-defected images. This allows for a soft clustering approach, making the model more robust compared to hard-assignment methods like K-Means.