Grounded-Segment-Anything icon indicating copy to clipboard operation
Grounded-Segment-Anything copied to clipboard

Interacting with GPT

Open pixeli99 opened this issue 2 years ago • 2 comments

How about using GPT as a super glue to achieve automated data annotation and training tasks, as mentioned in the readme, "A whole pipeline to automatically label images (with bounding box and mask) and generate new images."

Is there anyone interested? and what are your thoughts on this? I believe it's essential to have a broad discussion before getting started.

pixeli99 avatar Apr 10 '23 06:04 pixeli99

I think it's easier to realize such a system by adding tool prompts in VisualChatGPT or MMReact, but it may need some time to design the prompt. Have a good try!

Andy1621 avatar Apr 11 '23 02:04 Andy1621

@pixeli99 I have added a chatbot for our project. Please have a try!

Andy1621 avatar Apr 14 '23 00:04 Andy1621