Jinpeng Wang

Thanks for your attention on this work, I'll update this file latter.

Have you install the env?

Hi tankwin08: Follow previous install guide https://github.com/showlab/Image2Paragraph/blob/4c7f4436c02f5e9bc878fda0e352aa42995b3e07/install.sh

How about "pip install --upgrade pip". Because 1.9.0+cu111 is indeed in https://download.pytorch.org/whl/torch_stable.html, as below Or what's your system? I only try on linux If you have no GPU available, just...

Yes, the Mac install is different from linux. Look this page for instruction. https://pytorch.org

get it, thanks

It should be work now, thanks for report the bug.

Thanks for your suggestion. It would be easy and I'll consider implement it recently.

The SAM itself is not heavy. But semantic segment anything requires four large model which is very memory consuming. At now, simply use --semantic_segment_device as 'CPU' to run. We are...

Hi, we have implement a light version. Can be run on 8G GPU less than 20s.