Training large images for Image Classification task by creating BlockedImageDatastore

I am trying to train a CNN model using Deep Neural Designer App in Matlab. My images are large size images (i.e., 3000*1000) which I am to classify in three categories. Images are located in three different folders named according their class. However, GPU memory did not support when training the large images. I don't want to downsample the images as I may loose information. Now I am trying to see whether I can use blockedImageDatastore method to read the images in blocks (along with their classes). I couldn't find a good example for doing that. I would highly appreciate any guidance.

Answer 1

[Aylin]

Hello Zia, there's a couple of examples of using BlockedImageDatastore to train networks available here: link