GUIDEHow to Build AI Benchmarks that Evolve with your Model
Quick Start
Demo

Annotation UI Update

Label Studio Enterprise has a new annotations UI that’s more ergonomic, efficient, and ready to support larger, more complex tasks with dozens to hundreds of regions. The new UI features sleek new interactions and capabilities for annotators, including:

- New, dual ‘Region’ and ‘Details’ control panels vs a single sidebar, allowing you to view all pertinent information about the task without scrolling

- Collapsable, draggable and resize-able panels to customize your workspace, especially if you want to view images or video at a larger scale

- More granular controls for each region, including the ability to precisely set width, height and coordinates to fit an object for pixel perfect annotations

- Cleaner interface to group, organize regions, link, add metadata, and delete regions, plus new functionality to lock regions

Transcript

Hi, I'm Ilia, the front-end lead at Heartex, and today I'm introducing a new Label Studio feature called the Outliner.

When you open a task, you’ll notice a new two-column interface. You might remember the old single-sidebar layout—this new Outliner replaces that setup. It’s split into two panels: on the left, a list of regions that you can group or sort in different ways; on the right, a panel called “Details.”

Let’s walk through it. When you add a region to the image, it appears on the left. Clicking on that region brings up detailed information on the right. You’ll see exact coordinates for pixel-perfect annotation, along with tools for creating relations, adding metadata, hiding the region, deleting it, and a new feature—locking the region. When a region is locked, it becomes uninteractive and protected from changes.

You can also use quick actions in the Outliner to toggle visibility or lock status directly. For pixel-perfect annotation, you can edit coordinates manually and even rotate the region in the Details panel.

The Outliner is also flexible. You can detach the region list or details panel, move them anywhere on the screen, or hide them entirely to maximize workspace. You can also swap the left and right panels or resize them as needed.

Additional features include manual grouping of regions into a tree structure, grouping by label, or by the tool used (like rectangles, ellipses, or keypoints). If your regions were generated by a machine learning model, you can sort them by score—so higher-scored predictions appear at the top.

At the moment, the pixel-perfect annotation mode and the new Details panel are only available for images, but support for other data types is coming soon.

That’s the Outliner! We hope you enjoy using it—thanks for your time!

Related Content