search

jiyt17 / IDA-VLM

IDA-VLM: Towards Movie Understanding via ID-Aware Large Vision-Language Model
23 stars 0 forks source link

How to distinguish between ID image and test image #3

Open CserDu opened 1 week ago

CserDu commented 1 week ago

I downloaded and unzipped the beta_gpt4v_mix_mini_new.json file, but found that the id images and test images in each training sample were not marked, and I could not determine which images in the use instruction were id images and which were test images.

jiyt17 commented 1 week ago

The format of ID image in user instruction is "xxx is \<img>path\<\/img>", while test images are in other formats. Therefore, you can use regulations to judge.