is there any problem with function normalize in img_utils.py

[ghost]

def normalize(img, mean, std):
# pytorch pretrained model need the input range: 0-1
img = img - mean
img = img.astype(np.float32) / 255.0
# img = img / std

return img

hi ycszen, i feel that there is a problem about the normalize function, the img passed into the function is lied in 0-255, while the mean found in the config.py is typically 0.4-0.5, so if the normalize should be changed into

def normalize(img, mean, std):
# pytorch pretrained model need the input range: 0-1

img = img.astype(np.float32) / 255.0
img = img - mean
# img = img / std

return img

thank you very much

[yu-changqian]

Yes. You are right. It's my fault with the careless uploading. I have corrected it.