efficientnet on high resolution images

[RituYadav92]

can we use efficientnet as a feature extractor on high-resolution image input(1024x1024)? It will affect the accuracy or we should expect similar level results?

I tried the same but facing high validation loss. What could possibly be the reason behind this? Thanks!

[Yannick947]

Same question here, according to this link: https://tfhub.dev/google/collections/efficientnet/1. The image input sizes should be between 224 and 600 but what to do when my images are bigger and I don't want to rescale them because I need the details to distinguish between classes.

Thanks

[IngoDbg]

Did you find an answer to your question? I'm stuck with the same question right now.

Thanks