search

BR-IDL / PaddleViT

:robot: PaddleViT: State-of-the-art Visual Transformer and MLP Models for PaddlePaddle 2.0+
https://github.com/BR-IDL/PaddleViT
Apache License 2.0
1.22k stars 318 forks source link

[Semantic Segmentation] BEiT-L #50

Closed rock4you closed 2 years ago

rock4you commented 2 years ago

Does this model reproduced so far ? BEiT-L (ViT+UperNet, ImageNet-22k pretrain) mIoU-57.0% https://paperswithcode.com/paper/beit-bert-pre-training-of-image-transformers

xperzy commented 2 years ago

@GuoQuanhao @wutianyiRosun Could you guys help evaluate if we can add this model?

wutianyiRosun commented 2 years ago

@rock4you We are planning to reproduce BEiT-L.