Deep bacteria: robust deep learning data augmentation design for limited bacterial colony dataset

Bacterial colony classification is an important problem in microbiology. With the advances in computer-aided software's, similar problems have been solved in a speedy and accurate manner during the last decade. In this paper, deep neural network architecture will be presented to solve the bacterial colony classification problem. In addition, the training and testing strategy that relies on the strong use of data augmentation will be introduced. The used dataset was limited as it contains 660 images for 33 classes of a bacterial colony. Any neural network cannot learn from this data directly and in case of learning the neural network will overfit. The adopted training and testing strategy lead to a significant improvement in the training and testing phases. It raised the dataset images to 6,600 images for the training phase and 5,940 images for verification phase. The proposed neural network with the adopted augmentation techniques achieved 98.22% in testing accuracy. A comparative result is presented, and the testing accuracy was compared with those of other related works. The proposed architecture outperformed the other related works in terms of its testing accuracy.