IBM Developer Blog

Follow the latest happenings with IBM Developer and stay in the know.

New open source tool makes labeling data easier

Faced with the daunting task of hand labeling thousands of images, developers are looking for an easier way to train their object detection models. Currently, it takes 200-500 samples of hand-labeled images for a model to detect one specific object. Autolabeling images speeds the process and gives developers back valuable time to work on other innovative projects.

A new automated labeling tool in the open source Cloud Annotations project makes it easier for developers to label data without having to manually draw labels on an entire dataset of images. The new feature takes advantage of AI to assist in the labeling process.

You can access the tool now when using the Cloud Annotations GUI. You simply have to select the “Auto label” button and it will autolabel the images you’re uploading.

Cloud Annotations is a fast, easy, and collaborative open source image annotation tool. Backed by IBM Cloud Object Storage, Cloud Annotations enables you to store as much data as you need, access the data from anywhere and share across multiple collaborators in real-time.

Cloud Annotations enables multiple people to collaborate in real-time, access the data from anywhere, as well as store and back up the data through IBM public cloud.

How it works

To take advantage of the new tool:

  1. A user uploads and labels a subset of photos via the Cloud Annotations GUI.
  2. The user then trains a model following these instructions.
  3. The tool uses that model to label more photos. Simply select “Auto label” in the GUI and the tool will take over the task of labeling photos.
  4. User reviews new labels.

See it in action

The following video shows you how the autolabeling function works.

Get started

The Cloud Annotation auto labeling feature is currently live on GitHub, available for anyone to use and take advantage of as a massive time saver. Start using the tool and let us know what you think!