Deep- Image: Automated Identification of Bacteria based on Deep Learning Model

Zainab N. Al-Qudsy; Wasan Maddah  Alaluosi; Maad M.  Mijwil; Ahmed Adnan  Hadi; Mohammad  Aljanabi

doi:10.37934/ard.136.1.207220

Authors

Zainab N. Al-Qudsy Department of Intelligent Medical Systems, University of Information Technology and Communications, Biomedical Informatics College Baghdad, Iraq
Wasan Maddah Alaluosi Ministry of Education, Baghdad, Iraq
Maad M. Mijwil College of Administration and Economics, Al-Iraqia University, Baghdad, Iraq
Ahmed Adnan Hadi Intelligent Medical Systems Department, College of Sciences, Al-Mustaqbal University, Babil, Iraq
Mohammad Aljanabi Deputy Dean of Technical College, Imam Ja'afar Al-Sadiq University, Baghdad, Iraq

DOI:

https://doi.org/10.37934/ard.136.1.207220

Keywords:

ResNet, Bacteria images, Convolution neural network, Classification, Deep learning

Abstract

Accurate Classification of bacteria plays a crucial role in microbiology and beyond. It helps to identify infectious agents during epidemiological investigations, food safety monitoring, and detection of biological threat agents. Convolutional Neural Network (CNN) is a deep learning technique that has proven reliable in the field of Classification of medical and biological diseases. In this study, CNNs are utilized to develop a bacterial classification system. Within this system, Classification is subjected to several modifications before the ResNet method is used in order to identify the kinds of Bactria from among sixteen different classes of bacterial images. The model was fine-tuned by training only the last two layers of the pre-trained ResNet101V2 network, which significantly improved the performance. A large-scale dataset and confusion matrix were used to evaluate the model's performance. The experimental results demonstrate that the accuracy rate reached a peak of 98.66%. Moreover, the suggested approach enhances the advancement of automated diagnostic tools for bacterial pictures that surpass the present state-of-the-art models and provide the groundwork for future enhancements in bacterial image classification utilizing CNNs.