search

yuhuan-wu / P2T

[TPAMI22] Pyramid Pooling Transformer for Scene Understanding
200 stars 18 forks source link

weights #6

Closed just1226 closed 2 years ago

just1226 commented 2 years ago

我应该如何导入预训练好的权重到模型中？以便开始训练我的模型。

yuhuan-wu commented 2 years ago

Good question. P2T has four stages, each stage outputting one feature with scale of 1/4, 1/8, 1/16, and 1/32. You can add these features to a list (Like: https://github.com/yuhuan-wu/P2T/blob/8f3e3b21a5742d375dc5857cce602441b8905771/segmentation/p2t.py#L304), and then you can construct a FPN/U-Net with the output of P2T backbone.

just1226 commented 2 years ago

Thank you very much. I have solved.

yuhuan-wu commented 2 years ago

Good to hear that~