megvii-model / SinglePathOneShot

MIT License
259 stars 53 forks source link

Use BGR tensor in data preprocessing #5

Closed ultmaster closed 4 years ago

ultmaster commented 4 years ago

Hi. Thanks for this great repo.

I'm interested in why you did convert to BGR tensor intentionally and keep the values between 0 - 255 instead of normalization (div 255, minus mean, div std) as usual.

Related code:

https://github.com/megvii-model/SinglePathOneShot/blob/36eed6cf083497ffa9cfe7b8da25bb0b6ba5a452/src/Supernet/train.py#L37

ZichaoGuo commented 4 years ago

Just in order to keep the same as that in our own DL framework.

ultmaster commented 4 years ago

Thanks for response.