Image annotation tool

Hi everyone

Can anyone recommend a tool for annotating images for segmentation? I want to try out the Camvid notebook on my own data, so ideally the annotations should be downloadable in the same format as what we used in Lesson 3.

Regards
Ron

2 Likes

Hi,

How about this repo: https://github.com/wkentaro/labelme?

Labelme is a graphical image annotation tool inspired by http://labelme.csail.mit.edu.
It is written in Python and uses Qt for its graphical interface.

7 Likes

I haven’t tried it but CVAT looks great (although maybe complex to set up?).

There’s also image-labelling-tool which can apparently be used as a widget in a Jupyter notebook.

There are also closed source paid tools out there with a free trial.

Please let us know how you go

4 Likes

Thanks everyone - I’ll try them out and let you know!

Hi Ronald,

I have used label box before: https://labelbox.com. They provide the tool for free for smaller projects and I was extremely happy how easy it was to use and set up. Maybe it helps for your purpose as well.

4 Likes

There are a lot of tools to annotate images. A bit of context. Before the start of fast.ai course v3, I was attempting to replicate the 3rd place winner of the Kaggle DSB 2018 nuclei segmentation competition. The solution is the simplest (no ensemble), based on a single Mask-RCNN implementation by Matterport. While digging into this, I discovered this blog post:

From the blog post:

… I ended up using VIA (VGG Image Annotator) because of its simplicity. It’s a single HTML file that you download and open in a browser. Annotating the first few images was very slow, but once I got used to the user interface, I was annotating at around an object a minute.

If you don’t like the VIA tool, here is a list of the other tools I tested:

  • LabelMe: One of the most known tools. The UI was a bit too slow, though, especially when zooming in on large images.
  • RectLabel: Simple and easy to work with. Mac only.
  • LabelBox: Pretty good for larger labeling projects and has options for different types of labeling tasks.
  • VGG Image Annotator (VIA): Fast, light, and really well designed. This is the one I ended up using.
  • COCO UI: The tool used to annotate the COCO dataset.

Recently, I also stumbled upon Prodigy, from the maker of spaCy. I have not personally tested Prodigy, but looks very interesting.

12 Likes

Thanks Cedric!

For video check out Diffgram! https://diffgram.com/video