Skip to content
BLBlender

Blender Labelling Tool

by Denis Chera·Workflow

Images © Denis Chera. Used for product review; image policy.

Why we like it

Stands out in the superhive library for building video annotation directly on Blender's existing video, keyframing and collection systems, while staying free and open source under the MIT license.

What is Blender Labelling Tool?

Blender Labelling Tool is a free, open-source annotation add-on built on top of Blender 3 and later, released under the MIT license. It currently supports drawing bounding boxes on video and exporting them in YOLO format.

The workflow runs inside Blender: import a video, create a class collection, select it in the outliner, add a bounding box, then animate it before exporting. The tool draws on features Blender already provides, including support for multiple video formats, video proxies, different interpolation types, keyframing and the curve system, and the collection system.

Development has a roadmap. Short-term plans include COCO and Pascal VOC support, bounding box data import for YOLO, COCO and Pascal VOC, and 2D joint and skeleton data. Longer-term goals cover segmentation labelling, rotated bounding boxes, and 3D data labelling such as point clouds. You can download it from the developer's GitHub page, with an option to buy on Blender Market to support the team.

Who is Blender Labelling Tool best for?

Blender Labelling Tool is best for annotating bounding boxes on video inside Blender and exporting them in YOLO format.

How much does Blender Labelling Tool cost?

Blender Labelling Tool is completely free to download and use.

Key features

  • Annotate bounding boxes on video and export them in YOLO format
  • Runs inside Blender 3 and later as a standard add-on
  • Builds on Blender's own video handling: multiple formats, video proxies, interpolation types, keyframing, curve system and collections
  • Workflow of importing a video, creating a class collection, adding a bounding box and animating it
  • Free and open source under the MIT license
  • Installs from a ZIP archive through Blender's add-on preferences
  • Planned support for COCO and Pascal VOC formats

Tags

#annotation#bounding boxes#YOLO#machine learning#data labelling#video#open source

Frequently asked questions

What Blender versions does Blender Labelling Tool support?

Blender Labelling Tool is built on top of Blender 3 and later, and is listed for Blender versions 3.0, 3.1 and 3.2. Because it works as a standard Blender add-on, it relies on the host application's own video and animation systems. If you are running one of these Blender releases, you can install the add-on and begin annotating. For details on newer Blender versions, check the developer's GitHub page and the official documentation.

How do I install Blender Labelling Tool?

Blender Labelling Tool installs like any ordinary Blender add-on. First download the ZIP archive, then in Blender open Edit > Preferences, go to the Add-ons section and choose Install. Select the downloaded .zip archive, then enable the installed add-on from the list. Once enabled, the typical workflow is to import a video, create a class collection, select it in the outliner, add a bounding box and animate it before exporting. Further steps are covered in the official documentation.

Is Blender Labelling Tool free?

Blender Labelling Tool is listed as free here and is open source under the MIT license. You can download it from the developer's GitHub page at no cost. There is also an option to buy the add-on on Blender Market if you want to support the team behind it. Licensing and pricing can change over time, so check the developer's page for the current terms before relying on any particular arrangement.

What annotation formats does Blender Labelling Tool export?

Blender Labelling Tool currently exports bounding box annotations in YOLO format after you draw and animate them on a video. The roadmap adds more options over time: short-term plans include COCO and Pascal VOC support, along with bounding box data import for YOLO, COCO and Pascal VOC, and 2D joint and skeleton data. Longer-term goals extend to segmentation labelling, rotated bounding boxes and 3D data labelling such as point clouds. Check the developer's page for the latest on these features.

Similar plugins