Gaussian Noise Augmentation
Upon request we investigated if additive Gaussian Noise as form of data augmentation could yield any benefits for our baseline.
During training each patch is augmented with a new random sample of additive Gaussian Noise with zero-mean and standard deviation of 0.2. The same noise is applied to the validation data and also to the test images before testing.
This type of augmentation does not lead to improved results as shown in the following bar-plot.
Note: Baseline and DenoiSeg results are averaged over five runs and the error bars indicate one standard error about the mean. The baseline with additive Gaussian noise augmentation was only run once.