[Question] I have a question about capture to a custom data image.

[yspch2022]

Question

Hello, I'm making instruction-following data while looking at papers and reference documents using my custom images.
But I have a question.
When constructing instruction-following data, 'Context type 1: Captions' is used to make captions to describe the image,
how do I make captions for the image? I think it's made by putting it in a model such as BLIP, llava, or GPT4, right?