Fix __len__ and single_cls

[Laughing-q]

@AyushExel @glenn-jocher
related issues: #238, #229

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Updating label handling and dataset length reporting in YOLO data processing.

📊 Key Changes

• Ensured that class labels are consistently updated to zero for single-class datasets.
• Modified dataset length calculation to be based on the number of labels instead of the number of image files.

🎯 Purpose & Impact

• 🎨 The change to class label assignment ensures data correctness, which is crucial for model training accuracy in single-class scenarios.
• 📏 By deriving the dataset length from labels instead of image files, the code reflects a more accurate dataset size, improving dataset integrity and potential error handling in training loops.
• 🚀 These updates might lead to more robust model training performance, as they ensure that the dataset's structure and annotations are correctly interpreted by the training algorithm.

[github-actions]

CLA Assistant Lite bot All Contributors have signed the CLA. ✅

[Laughing-q]

I have read the CLA Document and I sign the CLA