classify/predict --save-txt fix

[glenn-jocher]

Signed-off-by: Glenn Jocher [email protected]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhanced output options for classification in classify/predict.py.

📊 Key Changes

• Added frame number to output paths for images and labels when processing video sequences.
• Improved image and label output for webcam streams or when processing multiple images.
• Changed the --save-txt flag default behavior to not save results to .txt by default.

🎯 Purpose & Impact

• Precise Tracking: Frame numbers in filenames help in tracking and comparing results over video sequences 🎞️.
• Flexibility: Adjustments ensure the code handles both single images and batch processing correctly, providing greater flexibility for different input types 🖼️.
• User Control: With the change to --save-txt, users now have more control over whether they want to save textual output or not, reducing unnecessary file generation 📁.

[glenn-jocher]

Classification --save-txt now saves one annotations file per image: