are the imgsize parameter of MixVisionTransformer is necessary？

lucidrains / segformer-pytorch

Implementation of Segformer, Attention + MLP neural network for segmentation, in Pytorch

MIT License

342 stars 43 forks source link

Closed AncientRemember closed 3 years ago

AncientRemember commented 3 years ago

segformer drop the position encoding and make the model inductive when test resolution is diff with train

lucidrains commented 3 years ago

@AncientRemember whoops, corrected! 🙏