Create labelstudio_vid_json2yolo.py

[Burhan-Q]

From Ultralytics Discord, request to include conversion from Labelstudio JSON to YOLO. This is the starting code shared with Discord user @hellodword who will be finishing work on this code.

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Introducing a new script for converting Label Studio video-frame annotations to the YOLO format. 🎥🎉

📊 Key Changes

• New Script Addition: A Python script (labelstudio_vid_json2yolo.py) has been added to convert annotations from Label Studio's JSON format into YOLO format, specifically for video frames.
• Annotation Processing: It processes different types of annotations, focusing on capturing the label and bounding box (bbox) details like center points (x, y) and dimensions (width, height).
• Frame-Specific Output: For each annotated video frame, it generates a corresponding YOLO-format text file, naming it with the video's name and frame number.

🎯 Purpose & Impact

• Purpose: To facilitate the use of video-frame annotations made in Label Studio for training models in YOLO, one of the most popular object detection architectures.
• Impact:
  • Enhanced Workflow: This script bridges the gap between annotation and model training, making it easier for researchers and developers to use Label Studio for video dataset preparation.
  • Accessibility: By automating the conversion process, it saves time and reduces errors, making the process accessible to a broader user base, not limited to those with deep technical expertise in data formatting. 🚀