Destroy Me: Automatic Artifact Generation for Histopathology Images
The results demonstrate that selective, impact-weighted augmentation is vital for balancing practical robustness with the preservation of subtle diagnostic features, and demonstrate that models trained on "destroyed" patches consistently outperform baselines on independent real-world datasets.
Zuzanna Krawczyk-Borysiak, Adam Krawczyk, Mateusz Miller et al.
· 0 citations