Closed karllandheer closed 6 days ago
Hi @karllandheer,
proper validation is something which unfortunately isn't always done, not only in the context of foundation models. I completely agree with you that they should compare against nnUNet (or any other task-specific baseline for that matter) and the comparison they are doing against SAM and MedSAM is also not really fair, as they work in completely different ways. MedSAM has a task specific nnUNet as a baseline in the publication.
Best, Yannick
Hello, I was wondering if anyone has done any comparisons of nnUNet with large scale foundation models like the one found here: https://arxiv.org/pdf/2405.12971
I find it strange they do not have bench mark comparison of task-specific nnUNet models? Perhaps someone in the community has investigated this?