Usage examples:

from utils.dataloaders import HUBDatasetStats

stats = HUBDatasetStats('coco128.yaml', autodownload=True)  # method 1
stats = HUBDatasetStats('path/to/coco128_with_yaml.zip')  # method 2

stats_json = stats.get_json(save=False)
im_dir = stats.process_images()

@kalenmike

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Refactored dataset statistics calculation and image processing in yolov5.

📊 Key Changes

• Introduced HUBDatasetStats class for dataset statistics and image processing.
• Moved various functions into the class as methods, including _unzip, _hub_ops, and _find_yaml.
• Added __init__ method to initialize class with paths and dataset verification.
• Added get_json method to compute and optionally save dataset statistics.
• Added process_images method to compress and save images for Ultralytics HUB.
• Streamlined logic by encapsulating functionality and improving code organization.

🎯 Purpose & Impact

• The new class-based approach organizes functionality better and simplifies the codebase.
• Provides a more maintainable and extendable structure for future enhancements.
• Impact on users includes potentially improved efficiency, reliability, and usability when dealing with dataset statistics and preparing images for Ultralytics HUB.