More model trained on SIN

rgeirhos / texture-vs-shape

Pre-trained models, data, code & materials from the paper "ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness" (ICLR 2019 Oral)

https://openreview.net/forum?id=Bygh9j09KX

Other

785 stars 101 forks source link

More model trained on SIN #8

Closed iminfine closed 5 years ago

iminfine commented 5 years ago

Hi Robert,

Thanks for your great works. It is useful for me.

I would like to implement the trained VGG-16 model on SIN to my work. Could you please provide it?

Bo.

iminfine commented 5 years ago

VGG-16 is not what I need, I am now training VGG-19 on SIN.