Handle large tiff files

Hi there all, I was wondering whether there is a fastai way to handle large tiff images, as the tiff images are huge and cannot be fed directly into the neural network model for training or to get rid of the white spaces in the images.


Please read here the section Berfore You Ask and summarise what you’ve tried from the first page of this search.

Hi @bencoman ,

Thanks for the post; it is not related to what I am looking for, but you are right to let me start by explaining what I am doing.

The code I am using is based on a kaggle notebook from IAfoss, 256x256 images this technique works well, but I was wondering if there are new approaches with the new fastai version that just came out.

Basically, I am exploring other possibilities of doing the same, but better as when you create the model to process these patches, it gets a little complicated.

Or a way not to split images in patches.

1 Like

While you are waiting for a better answer from someone more knowledgeable, you might consider:

So I presume you have actually tried using smaller images prescaled by a standard python non-fastai library. What are accuracies are you getting with those?

hi @bencoman,

Thanks for the links, but it is not what I am looking for.

1 Like

Does the new fastai version have methods/classes to deal with large TIFF files natively (during datablock declarations?)

1 Like

I found this great code from Jeremy Howard to resize images, it is part of the new FastaAI code.


1 Like

Another solution:

1 Like