Folder /Data/
contain two subfolders with images and their promts. Long-CLIP will be use test_image.jpg
to inference. If you want use your own pictures, save it to /Data/img/
as test_image.jpg
, then add new prompt to /Data/text/test_text.txt
- Clone this repository
- Download checkpoint LongCLIP-B and place it under
Long-CLIP/checkpoints/
- Build docker image
- Mount folder /Data to /app_volume and run image