NexumAI

NexumAI is a machine learning application that provides a user-friendly interface for training various types of models, including image classification, image segmentation, and voice classification.

Requirements

• Python 3.x
• PyQt6
• PyTorch
• TensorFlow
• pandas
• torchvision

Installation

1. Clone the repository:

git clone https://github.com/karami-mehdi/NexumAI.git
cd NexumAI

2. Install dependencies:

pip install -r requirements.txt

Usage

Run the application from the root directory:

python app/main.py

This will launch the GUI interface where you can:

1. Select your dataset
2. Choose the model type (Image Classification, Image Segmentation, or Voice Classification)
3. Select the specific architecture
4. Train and evaluate your model

Project Structure

NexumAI/
├── app/
│   └── main.py
├── core/
│   ├── model_generator.py
│   ├── model_saver.py
│   ├── model_selector.py
│   └── model_trainer.py
├── infrastructure/
│   └── data/
│       └── dataset_loader.py
├── presentation/
│   └── widgets/
│       ├── setting_window.py
│       └── trainer_widget.py
└── test/
    ├── test_model_generator.py
    ├── test_model_trainer.py
    └── uitest_trainer_widget.py

Model Types and Architectures

Note that some features are still in progress.

Image Classification

• CNN (Recommended)
• SVM
  • Alternative for smaller datasets
  • Traditional machine learning approach
  • In progress

Image Segmentation

• U-Net (Recommended)

  • Specialized architecture for semantic segmentation

• CNN

  • Basic segmentation capabilities
  • Suitable for simpler segmentation tasks

Voice Classification

• In progress

Testing

The project includes comprehensive unit tests and UI tests.

Model Generator Tests:

Model Trainer Tests:

UI Tests:

Accessibility

Light/Dark Mode

Font Size

Sequence Diagram

License

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