search

fuy34 / superpixel_fcn

[CVPR‘20] SpixelFCN: Superpixel Segmentation with Fully Convolutional Network
Other
392 stars 80 forks source link

How to set the mean and std when i want to change the training set? #32

Closed Surrin1999 closed 2 years ago

Surrin1999 commented 2 years ago

Hello, thanks to your nice work.

I am trying to improve your work. When I need to train your model on different dataset, how should the mean and std in the input_transform be adjusted? Or should it remain the same? image

Surrin1999 commented 2 years ago

The second mean and standard deviation (mean = [0.411,0.432,0.45], STD = [1,1,1]), I initially thought it was set according to the mean and std of BSDS500 training set, but after calculation, I found it was not.

How is this set?

fuy34 commented 2 years ago

Hey, I think you can set the mean and std of BSDS5000 as well.

This number will not affect the final result a lot. I directly copied the current value from the FlowNetPyTorch repo.

Surrin1999 commented 2 years ago

Thank you!😁