The best datasets aren't born, they're made.

DataTorch annotator, annotating a chess board with an annotation for each chess piece.

Great projects start with great data. DataTorch started with a great data annotator.

Data annotation is the cornerstone of our product. DataTorch was originally designed and developed by the creator of COCO Annotator, one of the most popular open-source annotation platforms in the world. Enjoy an annotation experience refined by hundreds of hours of feedback and iteration.

Plants with their fruits annotated.

The universal annotator.

Annotate images, video, text, and more with a variety of annotation types, including bounding box, polygon, keypoints, and file metadata. Utilize AI assisted tools such as SuperPixel and DEXTR or input your own models with pipelines.

Man posing during yoga, keypoints annotate his form.

See how every label fits together.

Labels are annotation classes in DataTorch. Quickly create and organize labels with parent-child relationships. Visualize the organization of your labels in both tree and list views.

Hierarchy of labels based off of chess pieces.

Different views for different roles.

Reorder and resize windows in the annotator and save views based on profile and permissions. Make the interface as simple or as complex as you like, depending on your needs.

Annotator interface showing off customization of panels.

Streamline your annotation and review process.

Create jobs, which allow you to distribute annotation and validation work amongst team members. Set a number of pass throughs and validation checks before approval. Calculate consensus amongst your annotators and validators and facilitate discussion with team comments and metrics.

List of annotation tasks to be assigned to annotator staff.