A modified geometric transformation and mean-based class filtering in data augmentation pipelines, paired with an offline interpolation-specific pipeline, yields performance gains on medical image segmentation datasets by avoiding nearest-neighbor label and detail issues.
Molecular Biology of the Cell 25, 457–469
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
Extension of prior multiscale framework to heterotypic cancer invasion incorporating dynamic fiber redistribution in two-phase ECM and leading-edge MDE processes, with computational exploration of invasion patterns.
citing papers explorer
-
Beyond Nearest Neighbor Interpolation in Data Augmentation
A modified geometric transformation and mean-based class filtering in data augmentation pipelines, paired with an offline interpolation-specific pipeline, yields performance gains on medical image segmentation datasets by avoiding nearest-neighbor label and detail issues.
-
Multiscale dynamics of a heterotypic cancer cell population within a fibrous extracellular matrix
Extension of prior multiscale framework to heterotypic cancer invasion incorporating dynamic fiber redistribution in two-phase ECM and leading-edge MDE processes, with computational exploration of invasion patterns.
- Semi supervised GAN for smart microscopy, fast and data efficient cell cycle classification