liupei101 / PseMix

[IEEE TMI 2024] Pseudo-Bag Mixup Augmentation for Multiple Instance Learning-Based Whole Slide Image Classification
44 stars 3 forks source link

How do you evaluate during training? #8

Closed weiaicunzai closed 1 month ago

weiaicunzai commented 1 month ago

Thank you for your work. I was wondering whether you still use PseMix during testing, or if you simply read the WSI features and pass them to the model.

If you are indeed still using PseMix during testing(although I don’t think this is the case), how do you determine the ground truth label of a given slide when you are mixing two slides together using PseMix?

weiaicunzai commented 1 month ago

Just make sure

liupei101 commented 1 month ago

Hi, @weiaicunzai, thanks for your attention.

For clarification, our testing does not involve data mixing; we follow common settings in model performance evaluation.

weiaicunzai commented 1 month ago

Hi, @weiaicunzai, thanks for your attention.

For clarification, our testing does not involve data mixing; we follow common settings in model performance evaluation.

Thanks!