According to documentation to apply resize in a databunch you just pass the size option in the transform() function. You can only change the resize method to crop,squish.pad.
Also fastai uses default torchvision.transforms.Resize() which provides no anti-aliasing option
In my case I use fastai for an image matting problem so I want my images to be resized with anti-aliasing .Otherwise the img loses some quality and gets jagged
Is there any workaround for this ?