sam-hq
sam-hq copied to clipboard

Published 20 hours ago •

Reame
Issues

About the code of finetuning on the specific dataset

Open momo1986 opened this issue 2 years ago • 10 comments

Thanks for this siginificant and interesting work. Open-source work is helpful to the developer In the paper, the authors write "In particular, we use both global semantic context and local fine-grained features by fusing SAM’s mask decoder features with early and late feature maps from its ViT encoder. During training, we freeze the entire pre-trained SAM parameters, while only updating our HQ-Output Token, its associated three-layer MLPs, and a small feature fusion block.". Is there a demo to run the training code or should we reproduce them by ourselves?

Thanks & Regards! Momo

Jun 13 '23 07:06 momo1986

Hi, thanks for watching HQ-SAM. We are organizing the training/finetuning codes and will release it after it's ready. Please stay tuned!

Jun 13 '23 08:06 lkeab

Cool. Looking forward to the training code.

Jun 14 '23 11:06 unriddler

Hi, thanks for watching HQ-SAM. We are organizing the training/finetuning codes and will release it after it's ready. Please stay tuned!

looking forward~！

Jun 15 '23 00:06 xiawei20161308104

looking forward~！

Jun 19 '23 02:06 astro-fits

looking forward~！

Jun 21 '23 07:06 momo1986

Thanks, we plan to release it at the beginning of July.

Jun 22 '23 20:06 lkeab

looking forward~

Jun 28 '23 07:06 yahooo-m

Looking forward, fine-tuning on this powerful tool would be excited!

Jul 03 '23 08:07 alexwong-16

Thanks. It will be ready this week.

Jul 03 '23 15:07 lkeab

Hi, we released the training codes for SAM tuning.

Jul 05 '23 13:07 lkeab