Hi All,
I am trying to understand the batch_stats output … I have tried implementing lesson1-pets notebook and stuck with output of batch_stats()
data = ImageDataBunch.from_name_re(path_img, fnames, pat, ds_tfms=get_transforms(), size=224, bs=bs).normalize(imagenet_stats)
data.batch_stats() return a list with two tuples : [tensor([0.0929, 0.1421, 0.1532]), tensor([1.1515, 1.1281, 1.2612])]
What i understand from source code is, It is channel wise reduction on single batch of train data. Since my train set has only 3 channels , batch_stats() should ideally return only one tuple … So what is the second tuple ? How can I interpret the second tuple ?
Which reduction (Mean or median ) is used for channel wise reduction ?