While following an idea after watching lesson 3, I got carried away a little bit and started making an app for creating image segmentation datasets.
Although it’s still in early stages and a lot has to be done you can see the current version here:
https://image-labelling-app.onrender.com/app
Some plans for the application:
- simple to use, get started quickly
- do as much as possible in the browser, no logins required
- support more labelling options than just pixel-wise coloring
- a desktop version with direct read/write to the filesystem
- ability to connect a network that automatically gets trained and generates masks -> then you fix what’s wrong
Are you interested in such an app? Or are the existing ones sufficient? If there already is an app that does exactly what I have written I’d like to know about it.