2D Anomaly Detection

This project focuses on detecting anomalies in 2D data using advanced machine learning models. It leverages the power of PyTorch for model training and Flutter for creating a cross-platform application that allows users to interact with the anomaly detection system.

📊 Training the Model

I used the MVTec AD dataset, which is specifically designed for benchmarking anomaly detection methods. The dataset contains over 5,000 high-resolution images divided into 15 different object and texture categories. Each category has normal images and images with various defects, making it ideal for training and evaluating anomaly detection algorithms.

✨ Features

• Pre-trained Models: Access pre-trained models like models/carpet.pt for quick deployment.

• 📱 Cross-Platform Application: A Flutter-based application (anomaly_detection_app_flutter) for easy interaction with the model.

• 🌐 Server Backend: A Flask server (models/server.py) that serves the model predictions.

• 🖥️ Desktop Application: A Tkinter-based desktop application for local anomaly detection.

⚙️ Installation

To set up the project, follow these steps:

1. Clone the repository to your local machine.

2. Ensure you create a virtual environment first with python version 3.7.1 .

    # Create a new conda environment with Python 3.7.1
    conda create --name myenv python=3.7.1
   
    # Activate the conda environment
    conda activate myenv

3. Install the required Python dependencies by running:

   pip install -r requirements.txt 

4. Download the backbone and add the backbone location to ADD/models/hrnet/hrnet.py at line 12.

5. Download the pre-trained model and place it in the models folder.

6. Navigate to the anomaly_detection_app_flutter directory and run the following command to install Flutter dependencies for the mobile application:

   flutter pub get

🚀 Usage

📱 For Mobile Application

1. To start the server, navigate to the ADD directory and run:

   $env:KMP_DUPLICATE_LIB_OK="TRUE"
   python -m models.server

2. To run the Flutter application, navigate to the anomaly_detection_app_flutter directory and execute:

   flutter run

🖥️ For Desktop Application

1. To run the desktop application, navigate to the ADD directory and execute:

   $env:KMP_DUPLICATE_LIB_OK="TRUE"
   python -m models.test

📷 Screenshots

Mobile Application

Desktop Application

Server

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

💘 Acknowledgements

• Thanks to the Flutter and PyTorch communities for their invaluable resources.
• Special thanks to the CDO Project team for their pioneering work.