Abstract:
The main objective of our study was to determine the effectiveness of machine and deep learning in diagnosing the causative bacteria for various underlying diseases, in order to make sure the instant and prompt diagnosis of the disease. For achieving this purpose study was conducted after obtaining written consent from the health care providers concerned with the testing and diagnosing purpose. The procedure further included the examination of the slides under electron microscope and then after obtaining the images deep learning models were applied. Three models were compared for deep learning and the best one was deployed. The results were studied to assess the role and effectiveness of deep learning methods for identification of bacteria for early and exact diagnosis and timely treatment of different diseases as well as to reduce the chances of complications, which suggested that deep learning can be useful tool for finding the bacteria responsible for different diseases. In this particular study we have used dataset from the Kaggle and applied various pre-processing techniques using VGG16 and VGG19. In our research conclusion report that training of the data on ResNet50 proves promising speed as compared to the VGG16 and VGG 19 model. The accuracy of the models are ResNet50 98.4%, VGG16 89.2% and VGG19 99.2%. In the light of the above results we can conclude that VGG 19 have more promising results as compared to the other model.
