htcr / sam_road

Segment Anything Model for large-scale, vectorized road network extraction from aerial imagery. CVPRW 2024
https://arxiv.org/pdf/2403.16051.pdf
MIT License
150 stars 18 forks source link

Value of "pixel_mean" and "pixel_std" #7

Closed chenpengxin closed 6 months ago

chenpengxin commented 7 months ago

This is a great job!

I noticed the hard-coded value of "pixel_mean" and "pixel_std" at https://github.com/htcr/sam_road/blob/main/model.py#L227 . I wonder why the cityscale and spacenet share the same values? Could you please give a suggestion of how to set these two values for my customized dataset?

Thanks!

htcr commented 7 months ago

Thanks for your interest. These are the values used by SAM if I remembered correctly. I don't think they matter a lot as long as they are reasonable values for natural images.

htcr commented 6 months ago

Closing, feel free to reopen if need further discussion.