Back
New Feature
Update

ALR: Work with Dense Annotations Effortlessly 🗺️

Up to 17x speed improvement in Annotation Action with Accelerated Label Rendering (ALR). Check it out.
ALR: Work with Dense Annotations Effortlessly 🗺️
June 2, 2023
5
mins read

Working with densely annotated files - for example those with more than 1000 annotations in a single image - can be a challenge for even the most powerful hardware. With so many annotations to render, the workview canvas can become slow and unresponsive, making it difficult to add, edit, or review annotations efficiently.

To address this, we're excited to announce the release of Accelerated Label Rendering (ALR) - a significant update to V7 that will improve the rendering performance of the workview canvas.

Accelerated Label Rendering introduces several key improvements that will make annotating dense data more efficient - you can expect up to 17x speed improvement in Annotation Action.

Here's what’s new.

Multiple Canvas Layers

Accelerated Label Rendering now supports multiple canvas layers, including:

  • The Item Layer
  • Inactive Annotations Layer
  • Active Annotations Layer
  • Overlays

Previously, every change in the workview canvas required V7 to re-render all annotations, which could be time-consuming and computationally intensive. With ALR, each canvas layer has its own stack of responsibilities and is rendered individually, depending on the user's actions. For example, selecting an annotation changes it from inactive to active, meaning that only the active annotation layer needs to be re-rendered. This ensures that V7 is always performing optimally and that users can work with dense annotations more efficiently.

Viewport Rendering

Accelerated Label Rendering also enables V7 to render objects in the viewport rather than the whole canvas. This means that when users zoom in, only the objects in the viewport are rendered, rather than the entire canvas. This makes rendering more efficient and ensures that users can navigate through the content effortlessly, without any lag or delay.

Have a look at the example below - the canvas has over 3000 annotations.

The rendering speed is not scaling linearly to the number of annotation - it’s constant, no matter the number of annotations.

With ALR, users can expect faster rendering times, smoother interactions, and an overall improved experience. We're excited to see how this improvement will help our users work more efficiently and effectively with dense annotations. Please contact your Customer Success Manager if you’d like to learn more.

Related updates

Hugging Face Integration: Easier Way to Use External Models
New feature
Update
Hugging Face Integration: Easier Way to Use External Models
Learn how to incorporate Hugging Face models into your V7 data annotation projects. Discover a wide array of models that will help you save costs and build better AI.
September 20, 2023
5
mins read
See more ->
Multi-Planar Annotations: Explore 3D Voxel Masks
New feature
Update
Multi-Planar Annotations: Explore 3D Voxel Masks
V7 now allows you to seamlessly integrate volumetric masks, enabling detailed 3D reconstructions of regions of interest. With the capability to adjust voxels from any perspective, you can achieve unmatched consistency and precision in your annotations.
September 20, 2023
6
mins read
See more ->
Gain Control of Your Training Data
15,000+ ML engineers can’t be wrong