Inference speed of the model

[rose-jinyang]

Hello
How are you?
Thanks for contributing to this project.
Which device did u test your models on?

You did NOT explain the device specification in the paper.

[xieenze]

Hi, we use V100 gpu to test

[rose-jinyang]

Thanks for your reply.
Do u think that the model can yield 20 FPS on a mobile device (ex: snapdragon 662) if the input image size is small (ex: 256x256)?

[xieenze]

Good question. I didn't try to run it on a mobile device but I believe this is important.
Maybe you need to apply model compression technologies, such as pruning or quantize on the SegFormer-B0.

[rose-jinyang]

Got it, thanks