MLP in Segmentation about eanet

[MenghaoGuo]

Thanks for your attention.
In my view, 1x1 kernel conv is equal to Linear Layer.

[colorblank]

Thanks. Are there multiple layers in MLP, or just one like the code?

[RobinhoodKi]

Great work! Would you like to share the speed of inference on Cityscapes with 1024*2048 resolution input? Or make some comparison with the SOTA real-time semantic segmentation work like DDRNet, BiSegNetv2, etc?

[MenghaoGuo]

Good suggestions.
I will try to integrate this block into real-time semantic segmentation framework in the future.
Thanks for your kind advice.