Could you please provide the script to generate the image and text features for cc3m?

[ilovecv]

Hi,

In current codebase, there is only feature2webdataset.py to convert image/text features to webdataset. Could you please provide the script to generate the image/text features for cc3m? Thanks!

[yuntaodu]

Hi, is there any updated?

[minimini-1]

https://github.com/baofff/U-ViT/blob/main/scripts/extract_mscoco_feature.py

I think they used this script.