Convert Datumaro annotation data to YOLO single image JSON format
• Structure: items (each with image and annotations), nested categories.label.labels
• Geometry: Supports points (polygons) and bbox (bounding boxes)
• Categories: Nested structure with categories.label.labels array containing label definitions
• Label References: Annotations use label_id to reference category indices
• height, width: Image dimensions
• key: Image filename
• boxes: Array of bounding box objects with label, x, y, width, height, confidence
• Bounding box format: Center coordinates (x, y) with width and height in pixels
• Labels are stored as strings (class names)
• Confidence field for future use (currently null)
• Datumaro → YOLO Single Image: Converts polygon annotations to center-based bounding boxes
• Polygon Preservation: Converts polygon points to bounding box coordinates
• Label Mapping: Preserves annotation labels from Datumaro categories
• Multiple Images: Each image gets its own YOLO single image JSON file
• Coordinate Transformation: Converts from top-left to center-based coordinates