YoloDatasetLab

YoloDatasetLab is an open-source Python package designed to simplify working with YOLO-based datasets. It provides comprehensive tools for collecting, converting, merging, and testing image, video, and frame-based datasets with YOLO models.

🚀 Features

• Easily convert and merge datasets in different formats
• Collect and process video, image, and frame-based data
• Easy testing and evaluation with YOLO models
• Command-line and PyQt6-based GUI support
• Open source and community-driven

📦 Installation

Clone the project:

git clone https://github.com/rag0nn/spiral_yolo_data_laboratory.git
cd spiral_yolo_data_laboratory

Alternatively, add the YoloDatasetLab folder to your project and import modules directly.

📘 Usage

• Manual usage:

  from YoloDatasetLab.tools.dataset import Dataset
  from YoloDatasetLab.tools.project import Project
  
  ds = Dataset("/path/to/dataset")
  ds.resize_images((640, 480))

• Terminal menu (Recommended):

  python YoloDatasetLab/main_terminal.py

• GUI (Work in progress 🚧):

  python YoloDatasetLab/main_gui.py

🛠️ Terminal Operations Functions

Below are the main functions in operations_terminal.py with brief descriptions:

EditOperations (Modify Data)

• create_project: Create a new project.
• create_dataset: Add a new dataset to the selected project.
• merge_datasets: Merge datasets in the project.
• copy_dataset: Copy a dataset to a specified directory.
• copy_dataset_to_same_folder: Copy a dataset within the same folder.
• import_from_same_folder: Import a dataset from the same folder.
• resize_images: Resize all images in the dataset.
• standart_split: Split the dataset into train/val/test with standard ratios.
• balanced_split: Split the dataset with class balance.
• create_sub_dataset: Create a subset of the dataset.
• rename_datas_consecutively: Rename files in the dataset consecutively.
• remove_data: Remove data from the dataset (currently inactive).
• export_unmatches: Export unmatched images and labels.
• convert_annotations: Convert annotations using a given mapping.
• apply_filters: Apply filters to images.
• slice_images: Slice images into parts.

ReviewOperations (Read-Only)

• switch_project: Change the active project.
• project_analysis: Show general project statistics.
• dataset_analysis: Detailed analysis of the selected dataset.

ModelOperations (Model Related)

• model_evaluation: Evaluate the selected model on dataset(s).

📂 File and Folder Structure

YoloDatasetLab/
  ├── __init__.py
  ├── config.yaml
  ├── main_terminal.py (For terminal control)
  ├── main_gui.py (For GUI control, !!! In Development 🚧)
  ├── operations_terminal.py 
  ├── operations_gui.py 
  ├── tools/
  │     ├── __init__.py
  │     ├── project.py (Project)
  │     ├── dataset.py (Dataset)
  │     ├── data.py (Holds data, image, and label)
  │     ├── enums.py (Constant Variables)
  │     ├── image_filters.py (Image Effect Filters)
  │     ├── reports.py (Data Reporting Classes)
  │     ├── object.py (YOLO Object Classes)
  │     ├── utils.py
  │     └── gui/
  │           ├── __init__.py
  │           ├── win_review.py
  │           ├── constants.py
  │           └── ... (other GUI files)
  ├── models/
  │     ├── ... (YOLO models and weight files)
  ├── test/
  │     ├── data/
  │     └── apply.py
  ├── projects/
  │     ├── ... (created projects)
  │     └── project_example
  │           ├── archive.py (Old files related to the project are stored here)
  │           ├── output.py (Statistics, data merging outputs are here)
  │           └── datasets (Datasets in the project are stored here)
  │                   └── dataset_example (Datasets in the project are stored here)
  │                         └── output
  │                         └── detect
  │                               └── images
  │                                   └── train
  │                                   └── test
  │                                   └── val
  │                               └── labels
  │                                   └── train
  │                                   └── test
  │                                   └── val
  │                               └── detect.yaml
  ├── LICENSE
  └── README.md

🤝 Contributing

We welcome your contributions! You can report bugs, add new features, or improve documentation.

1. Fork the repository
2. Create a new branch (git checkout -b feature/newfeature)
3. Commit your changes (git commit -am 'Description')
4. Push your branch (git push origin feature/newfeature)
5. Open a pull request

📝 License

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

---

Thank you for your suggestions, contributions, and feedback!
For more information and updates, visit our GitHub page.

---

Supported Files

.pt for nn models .jpg for image data .txt for annotation data