search

TIO-IKIM / CellViT

CellViT: Vision Transformers for Precise Cell Segmentation and Classification
https://doi.org/10.1016/j.media.2024.103143
Other
236 stars 41 forks source link

three-fold cross-validation #51

Closed CYYJL closed 3 months ago

CYYJL commented 3 months ago

Hi, Could you please explain how to perform three-fold cross-validation in your code? Isn't it about alternately using three datasets for training, testing, and validation?

FabianHoerst commented 3 months ago

Hi,

for three folds, you would need the respective config files, one your each fold. In the config files, you define the train and validation folds. The test fold is not regarded in the config file, rather there are specific evaluation scripts in which the test fold is then used.

Pannuke has the following split strategy: Training: Fold 1; Validation: Fold 2; Testing: Fold 3 Training: Fold 2; Validation: Fold 1; Testing: Fold 3 Training: Fold 3; Validation: Fold 2; Testing: Fold 1 See here: https://warwick.ac.uk/fac/cross_fac/tia/data/pannuke