Would you release the leading model and code for SCOD with Swin-L and Pre-trained on ImageNet-22K + Objects365 about egovideo

[cg1177]

Hello. We would release Swin-L finetuned on the SCOD task. But the one pretrained on Objects365 requires approval.

[bryanzhou008]

Thank you very much, look forward to your release!

[bryanzhou008]

Hi there @cg1177, any updates to the model release? Thanks in advance!

[czczup]

Hi there @cg1177, any updates to the model release? Thanks in advance!

I will upload this model today.

[bryanzhou008]

Hi there @cg1177, any updates to the model release? Thanks in advance!

I will upload this model today.

Thanks a lot!

[bryanzhou008]

@czczup @cg1177 A quick follow-up question: could you state the per GPU memory required by the training script for the large model (or just the GPU type/per GPU memory used in this work)? Thanks again!