How to extract correct coordinates from the image based on feedback
so im working on a project which is AI Assisted Design feedback web app, im sending a design image to the backend, open ai gpt 4 analyses that image and sends back feedback and the coordinates in form of rectangle which encircle the part of the image the feedback is about, but the issue is the ai is not able to give proper coordinates about the part of the image that it is giving feedback about, how can i fix this..?
i want to highlight parts of image, for which im using coordinates, but it seems the open ai hallucinates and isnt able to give proper coordinates for this
would love to know if theres a way to fix this problem
1 Reply
Can you ask AI to describe the part of the image then use a different generative model to add a circle to the existing image? (I have not tried this idea)