Segmentation and genus classification of lactic acid bacteria from images

This comprises the full dataset used in the project "Segmentation and genus classification of lactic acid bacteria from images". It includes a training set, validation set and several test sets. Lastly it includes backgrounds which are needed for the custom image augmentation method described in the manuscript. The training and validation set are comprised of 1360 grayscale phase-contrast images containing >760,000 lactic acid bacteria which have been hand-labelled for segmentation (pixel classification) and classification of bacterial genus. Each image in the training and validation set contains only a single genus is named according to what genus it contains, e.g. "Lc" for Lactobacillus.