cxr8

cxr8 copied to clipboard

Question

4

Hey TRKuan, it is not really an issue, it is more that I want to classify the no findings from the sick people and wanted to make sure I understand...

Shaam93

Have you successfully trained the network?

8

Hi TRKuan, This is a good repo! Have you successfully trained the network? I tried the training code you provided, the loss is almost 0 and the AUC is around...

tangyuxing

https://github.com/TRKuan/cxr8/blob/77ad689c9364bceb791c7d913fa429b9e3413030/dataset.py#L125 "RuntimeError: invalid argument 2: size '[512 x 512 x 1]' is invalid for input with 1048576 elements at /pytorch/aten/src/TH/THStorage.c:41"

EyalRozenberg1

Hi, I have a question about the weights in the loss functions. Shouldn't the weights be different per class? If I understood well your code, you summed up all 1s...

YangLiu-belgium