Image labeling and segmentation tool2024-04-05T10:29:48+02:00

Annotate images fast and accurately with ML-powered tooling.

  • Use 3D labels to label 2D images faster
  • Access smart labeling tools for unseen speed and accuracy
  • Label data faster by leveraging your model

Trusted by computer vision engineers across robotics and AV companies

Label 2D images & sequences quickly

Label objects and regions quickly and accurately. gives you access to ML-assisted automation tools like Superpixel 2.0 and Autosegment.

We support:

  • Segmentation (instance, semantic and panoptic)
  • Bounding boxes
  • Keypoints
  • Polygons
  • Polylines’s Superpixel tool for image segmentation boosts our labeling efficiency significantly. The workflows, Python SDK, and webhooks have been game changers for us. The platform is now seamlessly integrated with our active learning pipeline.

Andres Milioto

Senior Computer Vision Engineer
Author of SemanticKITTI and RangeNet++


Get more precise image segmentation

Superpixel 2.0 takes the traditional Superpixel approach and uses ML models to recognize shapes, rather than relying on the color of nearby pixels.

Perfect for autonomous driving and robotics use cases, the models will give you more precise and accurate segmentation of images.


Automatically segment small and fine objects

When you need to segment small objects or high resolution images, Autosegment is here to help.

Draw a box around the object you want to label, and Autosegment will generate a segmentation mask for the foreground element in the box. It’s a perfect complement to the Superpixel technology.


Segment images with a single click

Built upon a massive dataset of over 11 million images, the Segment Anything Model (SAM) integration lets you segment any object in a single click.

Just hover your mouse over the image to view the suggested segmentation masks, then click the one you want.

Model assisted labeling workflow

Increase speed with your own ML models

Save hours in the data labeling process and improve the quality of your labels at the same time.

Upload your model predictions to create new labels, then use smart labeling tools to correct any errors in the predictions.


Choose the label type that works best for your project

Choose whether you want to label your images using segmentation masks, vector labels, or bounding boxes — and how you want to view and export your labeled data. lets you easily switch between the instance or semantic view of your labeled data so you can see it from different perspectives. You can also export the data as instance, semantic, or panoptic segmentation labels.

Need a different export format? supports multiple common export formats, allowing you to integrate your labeled data with other tools and platforms easily.

Semantic segmentation for autonomous driving images

smart interpolation with track id for image data labeling

Track objects across frames

Easily track an object’s movements and actions, even as it moves in and out of the frame or changes its appearance. lets you assign a unique identifier to each object in a sequence, then tracks it across frames so you have a record of its movements.


Limit the number of bounding boxes you have to draw

Speed up bounding box labeling by automatically interpolating box position and dimensions between keyframes.

Skip a few frames before editing your bounding box, and will automatically update the annotation in the missing frames for you.

image sequence video smart interpolation with track id for image data labeling

There’s a lot more that you can do helps machine learning teams and data engineers move from POC
to production faster, and at scale. Get in touch today to hear about all the neat labeling features available.

Customizable export2023-10-24T21:53:54+02:00

To download your labeled data, create a Release on the Releases tab of your dataset. A Release is a snapshot of your dataset at a specific point in time.

By clicking the download link of a Release, you’ll get a Release file in JSON format. This Release file contains all information about the dataset, tasks, samples, and labels in the release.

Using the Python SDK, the Release file can be exported to a variety of common formats:

  • COCO-instance segmentation format
  • COCO-panoptic segmentation format
  • YOLO Darknet object detection format
  • Grayscale PNGs (16-bit) where the values correspond to instance ids
  • Colored PNGs where the colors correspond to different instances
  • Colored PNGs where the colors correspond to different categories, with colors as configured in the label editor settings when available
  • For exporting segmentation bitmap labels to polygons
Customizable hotkeys2023-10-24T21:33:50+02:00

Make labeling even faster and more efficient with customizable hotkeys. You can use hotkey shortcuts to:

Rotate, pan, move, and scroll
Zoom in on a view
Define rotation points for an object
Switch to an orthographic top view
Create and delete cuboids.

For more information on hotkeys, check out the documentation.

Dataset search syntax2023-10-24T21:34:47+02:00 makes it easy to find samples, fast. You can search for samples by name, metadata attributes, or label content.

You can learn more about the data search syntax in this documentation.

See object counts from the objects sidebar2023-07-03T17:23:12+02:00

With the search bar, you can search for samples by their name, metadata attributes, and label content.

Add image and frame attributes, and make them optional or mandatory2023-10-24T21:54:57+02:00

Attributes come in 4 types:

  • Select box
  • Text
  • Number
  • Checkbox

You can set a default value for each attribute or make them mandatory.

In the label editor, the object-level and image-level attributes will be shown in the sidebar on the right. The image-level attributes are always visible, while the object-level attributes are only shown when an object is selected and has a category with object-level attributes.

Annotation types2023-07-03T17:22:13+02:00
  • Semantic segmentation
  • Instance segmentation
  • Panoptic segmentation
  • Bounding box
  • Polygon
  • Polyline
  • Keypoint

Labeling on is so quick and simple. Plus, the documentation is very good. It does everything we need, so we don’t have to worry about it.

Libor Novak

Software Engineer

Labeling Services

If you’re ready to outsource your labeling, we’re here to help.

Reach out today, and we’ll connect you with one of our trustworthy labeling partners.

Picture of someone labeling multi-sensor data on a point cloud

A few of the labeling agencies we work with

Simplify multi-sensor labeling

Say hello to faster labeling, better quality data, and more time for your engineering team.
You’ve got access to a 14-day free trial to test it for yourself.

Go to Top