How many images are there after using transforms

vision
1

I have a special dataset and images in the dataset are with channels more than 3.

So I want to chage the code in torchvision.transforms. I do not know how many images will be produced after the transform.
For example, I input one image with the shape 256 256 30, and I want to crop the input image to size 224 224 30. How many images will produced normally for RGB images?

I also try to write a new crop function for numpy array, then combine it with transforms.Lambda, but when I use the transforms.ToTensor():
ValueError: some of the strides of a given numpy array are negative. This is currently not supported, but will be added in future releases.

Thanks!!

2

Adding transformations to your Dataset does not increase the number of images, but applies these transformation on each sample.
The number of samples is defined by the return value in Dataset.__len__.

E.g. if you are dealing with 1000 images, you’ll now get 1000 (randomly) transformed images.

There are a few exceptions, such as transforms.FiveCrop and transforms.TenCrop, which will yield 5 or 10 crops, respectively, and thus increase the number of samples.