Creating a dataset for instance segmenation?

Going through the lesson 3, Jeremy used a pre-labled dataset for semantic segmentation.
I would like to create a custom dataset for instance segmenation.
It seems instance segmenation is even more accurate as they are bold curves around the boundaries?
Are there any tools I can use to create such a dataset?

Has anyone created such as dataset?

Thank you.

1 Like

The most popular tools are sloth, rectlabel, labelme, labelbox. They have different strengths/weaknesses - there is no ‘killer app’ yet in this space.

Currently I’m developing a new tool for this use case: image.predicted.ai