Welcome to YoloDataHelper 💡

YoloDataHelper is a small Python utility to process YOLO(you only look once) datasets. This is a utility tool for merging datasets, augmenting data, removing classes, visualizing annotations, and other operations that make working with YOLO datasets easier by developers and researchers.

🛠️ Features //

1. Dataset Combination

• Combine multiple YOLO datasets while properly aligning classes and adjusting label IDs.
• Retain the original structure of the datasets and generate a unified data.yaml file.

2. Data Augmentation

• Apply various transformations to YOLO dataset images, such as:
  • Hue, saturation, and brightness adjustments.
  • Contrast enhancement.
  • Adding random noise.
  • Color jittering.
• Generate augmented images with updated labels.

3. Class Removal

• Remove specific classes from the dataset and their associated images and labels.
• Automatically adjust class IDs and update the data.yaml file accordingly.

4. Annotation Visualization

• Display bounding boxes or segmentation masks over images for easy verification.
• Save annotated images to a specified output directory.

5. Classes Equalization

• Balance the number of images per class to ensure a uniform distribution.
• Adjust the dataset to prevent class imbalance issues.

6. Dataset Validation

• Ensure the presence of the necessary directories (train, valid, test) and their subfolders (images, labels).
• Automatically create any missing directories if they don’t exist.

7. Resize Options

• Compression: Resize with compressing images
• Advanced_compression: Resize with advanced compressing images
• Crop: Resize with cropping images
• Advanced_crop: Resize with advanced cropping images

---

📦 Installation //

Clone the Repository

To get started, first clone the repository:

git clone https://github.com/alireza-py/YoloDataHelper.git
cd YoloDataHelper

Install Dependencies

Install the necessary dependencies using pip:

pip install -r requirements.txt

🚀 How to Use //

Run Directly To use the tool as a standalone application, simply run the main.py file:

python main.py

1. Dataset Combination

Combine multiple YOLO datasets into a unified dataset:

from YoloDatasetsTools import DatasetProcessor

datasets = ["path/to/dataset1", "path/to/dataset2"]
output_path = "path/to/combined_dataset"

processor = DatasetProcessor(output_path)
processor.combine_datasets(datasets)

2. Data Augmentation

Apply data augmentation to a dataset:

from YoloDatasetsTools import DatasetProcessor

output_path = "path/to/augmented_dataset"
augmentation_params = {
     'hue': (-10, 10),
     'saturation': (0.7, 1.3),
     'brightness': (0.7, 1.3),
     'contrast': (0.8, 1.2),
     'noise': (10, 50),
     'color_jitter': (0.9, 1.1)
}
processor = DatasetProcessor(output_path, augmentation_params=augmentation_params, multiplier=3)

processor.process_folder(input_folder="path/to/dataset")

4. Annotation Visualization

Visualize bounding boxes or segmentation masks:

from YoloDatasetsTools import DatasetProcessor

output_path = "path/to/visualized_dataset"
processor = DatasetProcessor(output_path)

processor.visualize_annotations(dataset_folder="path/to/dataset")

5. Classes Equalization

from YoloDatasetsTools import DatasetProcessor

cleaner = DatasetCleaner(dataset_folder="path/to/dataset")

cleaner.classes_equalization(subset=["train", "valid", "test"])

6. Directory Validation

Ensure required directories (train, valid, test) and their subfolders exist:

from YoloDatasetsTools import DatasetProcessor

dataset_path = "path/to/dataset"
processor = DatasetProcessor(dataset_path)

processor.ensure_dataset(dataset_path)

7. Resize Options

from YoloDatasetsTools import DatasetProcessor

dataset_path = "path/to/dataset"
output_path = "path/to/output_dataset"
size = (720, 340)
mode = "advance_crop"

processor = DatasetProcessor(dataset_path)

processor.process_resize_and_crop(
  dataset_path, 
  output_path, 
  size,
  mode
  )

📚 Directory Structure //

This tool assumes the following directory structure for YOLO datasets:

dataset/
  ├── train/
  │   ├── images/
  │   └── labels/
  ├── valid/
  │   ├── images/
  │   └── labels/
  └── test/
  │   ├── images/
  │   └── labels/
  └── data.yaml

The data.yaml file should include:

• train, val, and test: Paths to the respective datasets.
• nc: The number of classes.
• names: A list of class names.

💥 Contributing //

Contributions are welcome! If you'd like to contribute to YoloDataHelper, you can:

• Fork the repository.
• Create a new branch for your feature or bug fix.
• Commit your changes and push the branch.
• Open a pull request with a description of the changes.
• If you encounter any issues, feel free to open an issue in the repository.