mp_sdk.training.ImageAnnotation

 class mp_sdk.training.ImageAnnotation(*,
     id: int,
     label_id: int,
     contour: List[List[float]],
     bounding_box: Tuple
)

Image annotation object.

Represents a single annotation provided by the application user. Usually used with mp_sdk.training.AnnotatedImage object.

All coordinate points provided in the bounding_box and contour attributes are relative to the image size, meaning that the coordinates are in range from 0 to 1 representing the percentage of the image size.

Attributes

• id - int: The annotation ID, unique within AnnotatedImage.annotations list.

• label_id - int: The label ID refencing a label from TrainingSession.annotation_labels list.

• contour - List[List[float]]: Points of the annotation contour list of [(x, y)] tuples.

• bounding_box - Tuple[float, float, float, float]: The bounding box of the annotation with [min_x, min_y, max_x, max_y] values.

• yolov5_annotation - str: Annotation in YOLO format returned as a string.