The compressed image resolution affects to Deep learning accuracy?

[leesangjoon1]

Using detectron2, .yaml file compress the image size to the size what we want ( ex) real image size 5544 x 5544 -> the size what we want 4000 x 4000)

then. The image is compressed. then It affects to Deep learning accuracy?

And If my camera specification is very superior and so the pixel size is so high, the Deep learning accuracy will be better?

When the training is progressed, the image size would be decreased by convolution.

So anyway the image is decreased in training, I think the image resolution is not affecting to accuracy critically if we have proper resolution

Please reply this . Thank you!