Generating images of hydrated pollen grains using deep learning

Abstract Pollen grains dehydrate during their development and following departure from the host stigma. Since size shape of a pollen grain can be dependent on environmental conditions, being able to predict both these factors for hydrated dehydrated state could beneficial in fields climate science, agriculture, palynology. Here, we use deep learning transform images Ranunculus into grains. We also then neural network that was trained experimental different genera identify generated transformed images, test accuracy image generation network. This pilot work demonstrates first steps needed towards creating general learning-based rehydration model useful understanding predicting morphology.