Data Labeling Studio

Image Annotation

from labeling_studio import ImageAnnotator
Initialize annotator
annotator = ImageAnnotator(
    annotation_type="bounding_box",
    labels=["person", "car", "dog", "cat"],
    output_format="coco"
)
Annotate images with active learning
annotator.annotate(
    image_dir="./images",
    output_file="./annotations/coco.json",
    active_learning=True  # AI suggests uncertain samples
)
Export to YOLO format
annotator.export("./annotations", format="yolo")

Text Annotation

from labeling_studio import TextAnnotator
NER annotation
annotator = TextAnnotator(
    annotation_task="ner",
    labels=["PERSON", "ORG", "LOC", "DATE"]
)
Annotate from file
annotations = annotator.annotate(
    text_data="./data/corpus.txt",
    output_file="./annotations/ner.json"
)

Quality Check

from labeling_studio import QualityChecker
Check annotation quality
checker = QualityChecker()
report = checker.check(
    annotations="./annotations/coco.json",
    ground_truth="./annotations/ground_truth.json",
    metrics=["iou", "consistency", "coverage"]
)
print(f"Average IoU: {report['iou']:.2f}")
print(f"Consistency Score: {report['consistency']:.2f}")
print(f"Coverage: {report['coverage']:.2f}")