#19 · Primary category: Data Annotation & Labeling Tools

LabelMeAnnotationTool

annotation computer-vision

Source code for the LabelMe annotation tool.

Project last updated:03/29/23

GitHub Stars

1.5K

Forks

444

Contributors

19

License

MIT

Why we included this project

LabelMe is one of the older self-hosted annotation tools for computer vision, and it has held up well. You run it on your own web server, and annotators label images through a browser, so nobody needs to download your dataset or install software to help out. That makes it easy to spread the work across many people, and the tool even connects to Mechanical Turk if you want paid crowdsourcing. It saves polygon outlines and segmentation masks, plus a scribble mode for quicker marks, in plain folder structures that a training pipeline can read directly. That combination of self-hosting and browser access is a big reason it has been a standard reference in the vision community for well over a decade.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category