On the PyTorch documentation for torchvision.models, it is states that images have to be loaded in a range of [0,1] and then normalized using mean = [0.485, 0.456, 0.406] and std = [0.229, 0.224, 0.225]. What is the logic behind these specific values?
Topic torchvision pytorch deep-learning
Category Data Science
You can find the answer in stackoverflow here and here, and also in the pytorch discussion forums
As described in the links, these values were computed over the Imagenet training data. From the last link, which is the explanation by Soumith Chintala:
On Imagenet, we’ve done a pass on the dataset and calculated per-channel mean/std.
Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.