Annot8 - The fastest way to tag images for object detection datasets
I built Annot8 because I thought current options for labeling object detection datasets were way too slow.
With drag-and-drop uploads, hot-keys, instant export, it’s now much faster. If you're training vision models, I seriously think it'll save you time
Replies
findable.
very interesting. What's your goal with Annot8?
Leavoo
Coming with some research background in cv and imaging, I think this could be very useful for the scientific community and time saver. Good job. I think it could be also good idea to share it in other more scientifically focused communities, it may help.
Haplo AI
@youssef_abdelwahed Any suggestions on additional communities to post it in?
Leavoo
@jcass99 Nothing specific in my mind right now. Idea I would suggest, may be write an article on medium or other place...where you can describe the different ways for annotating the data, and bring the tool as one of the options...etc.
Congrats on the release. Is this something that you will be maintaining?
Haplo AI
@yellowpencil Absolutely! It's something I am using personally pretty frequently, so I will be pushing updates regularly!
Hotkeys + drag-and-drop sound like a dream! But does Annot8 support:
- AI-assisted pre-tagging (e.g., SAM/YOLO auto-suggestions)?
- Bulk-edit tools for correcting misclassified objects?
- Customizable pipelines (e.g., `if [class="cat"] → auto-adjust bbox padding`)?
My 50k-image dataset is begging for this! 🐱
GPT-4o
Dang, instant export is a life-saver—no more waiting around for files, just done and out, fr. Makers really nailed the speed thing imo!
I think It going to be very interesting and helpful in comming time.