Giter Site home page Giter Site logo

Automation of annotation about coco-annotator HOT 5 OPEN

carmenrt avatar carmenrt commented on June 2, 2024
Automation of annotation

from coco-annotator.

Comments (5)

SixK avatar SixK commented on June 2, 2024

If you already have a model that can find "things" you want to annotate, you can use a tool like "https://github.com/ozayr/detection-assisted-annotation-tool" that will generate coco yaml files of things detected.
Coco-annotator will let you to import generated files.
If you need to annotate manually, you can use integrated DEXTR model/tool, Maskrcnn model/tool or Magic Wand integrated tool to help to generate mask, but depending on image complexity, it may be faster to manually create mask.
Note that in forks you will be able to find other model/backend support (for exemple detectron2)

Also, I just published a SAM (Segment Anything) mask generator server here to use with coco-annotator:
https://github.com/SixK/sam2coco_server

But it's actually a basic Proof Of Concept, and you may have to modify it to fit your needs.

from coco-annotator.

carmenrt avatar carmenrt commented on June 2, 2024

I have microscopy images. I want to select each cell with differents IDs but with the same category. Can you explain me how I can use Magic Wand Tool in this case?

Thank you!

from coco-annotator.

SixK avatar SixK commented on June 2, 2024

Magic Wand work by floodfill (selection propagate to near colors around), you have create a new annotation (+ button on category) for each cell you want to label. It require a neat color change to stop selection properly. If not, you will have to finish selection by hand using brush, eraser, or polygon tool.

Once you have selected your new annotation.
click on magic wand tool.
click once and it will fill space or stay mouse button down and move your mouse if selection is not complete.
if you want to remove parts of selection, you can press Shift key while you click.
You can also click several times on Undo button to cancel latests selections

For better selection with Magic Wand, you will have to adjust Threshold and Blur in Magic Wand Panel (where you find Magic Wand under categories panel at right...).

So, it's Magic, but this tool have it's limitations and may not work fine in all cases

from coco-annotator.

carmenrt avatar carmenrt commented on June 2, 2024

Are there any video tutorials available on this tool?

from coco-annotator.

SixK avatar SixK commented on June 2, 2024

For Magic Wand tool, you can have a look at coco-annotator videos on Youtube (not sure any video explain this tool):
https://www.youtube.com/results?search_query=coco-annotator

For Segment Anything (SAM) you can have a look at this demo (There an example on microscope image):
https://segment-anything.com/demo

And if you want to try SAM in coco-annotator, I released a Vue3 port with integrated SAM (SAM-HQ in fact actually).
But you will have to build docker images (require probably at least 18Go space), and this version has few bugs (wrong id on masks if an annotation is not selected, javascript error when playing with keypoints).
Also, Generated coco annotations file are not verified yet
https://github.com/SixK/coco-annotator/tree/vue3+sam

from coco-annotator.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.