How about the performance using ImageNet pre-tranied ViT?

TIO-IKIM / CellViT

CellViT: Vision Transformers for Precise Cell Segmentation and Classification

https://doi.org/10.1016/j.media.2024.103143

Other

217 stars 33 forks source link

How about the performance using ImageNet pre-tranied ViT? #23

Closed windygooo closed 12 months ago

FabianHoerst commented 12 months ago

We have not checked the performance using ImageNet pre-trained weights. Good idea, we will probably perform experiments on that. We already updated the results when using STARDIST and CPP-Net model structure.