microscopy images; corruption simulation

      We have developed three datasets, referred to as ER-C, Mito-C and Nucleus-C, respectively, for benchmarking robustness of DNN models against corruptions and adversarial attacks in semantic segmentation of fluorescence microscopy images.  Degraded images in these three datasets are synthesized from raw images along with their manually annotated segmentation labels in the ER, Mito, and Nucleus datasets [1,2], respectively. They are synthesized with controlled corruptions and adversarial attacks.  An updated version of the dataset has been released at