Struggling to get the label/description of annoted objects in Image

Hi All,
I am trying to get the description of objects in Image annoted with bounding boxes using highlighted color.
Can someone please provide the guidance in this, how to achieve it?

1 Like

Welcome @gk646

You can use JSON mode on gpt-4-turbo.

Here’s one of many ways to do that:

Describe the JSON object you want it to reply with, and then include a sample of the annotation color(s) in the second system message using an image to help the model understand which colored boxes to respond to.

Additionally, you can also label the annotation boxes with numbers in case there are more than one.

4 Likes

Glad it solved your issue!

Hi @gk646

We appreciate your participation in our community but we’d like to emphasize that posts on our forum can be very helpful to others facing similar issues.

please refrain from making edits like

as this is seen as vandalism and removes useful information for others, If you need assistance with your post, please let us know, and we’re happy to help guide you through the correct process.