Normalization of input images

[zzg-zzg]

Dear authors,
I wonder if the input image data has been standardized preprocessed (e.g. mean_vals = [0.471, 0.448, 0.408] std_vals = [0.234, 0.239, 0.242]). I can't find the relevant code.
If the answer is no, I wonder if it would be beneficial to do so, since many pretrained models use standardization.
Thanks

[Sawyer]

I simply normalize the RGB value to [-1, 1] since I do not load the pre-trained weights trained on typical datasets for 2D vision tasks (e.g. Imagenet, COCO, etc). I think there will be no significant difference between the standardization patterns when training a model from scratch, especially not on a very large-scale dataset.

[zzg-zzg]

Thanks for your reply.