MedicalNet icon indicating copy to clipboard operation
MedicalNet copied to clipboard
verified publisher icon Tencent

Pre-trained models' results

Open mr-BM opened this issue 3 years ago • 0 comments

Pre-trained models' dice scores on the LungSeg dataset are mentioned two times in the Readme, but they differ. The results in https://github.com/Tencent/MedicalNet#update20190730 look suspicious (it is suprising that ResNet10 performed significantly better than ResNet50).

Which results are reliable?

mr-BM avatar Sep 09 '22 11:09 mr-BM