Why should we normalize images?

Normalization helps get data within a range and reduces the skewness which helps learn faster and better. Normalization can also tackle the diminishing and exploding gradients problems.

Image transformation is a process to change the original values of image pixels to a set of new values. One type of transformation that we do on images is to transform an image into a PyTorch tensor. When an image is transformed into a PyTorch tensor, the pixel values are scaled between 0.0 and 1.0. In PyTorch, this transformation can be done using torchvision.transforms.ToTensor(). It converts the PIL image with a pixel range of [0, 255] to a PyTorch FloatTensor of shape (C, H, W) with a range [0.0, 1.0]. 

The normalization of images is a very good practice when we work with deep neural networks. Normalizing the images means transforming the images into such values that the mean and standard deviation of the image become 0.0 and 1.0 respectively. To do this first the channel mean is subtracted from each input channel and then the result is divided by the channel standard deviation. 

output[channel] = (input[channel] - mean[channel]) / std[channel]