Design and Construction of a Digital Microscope for Automatically Counting Escherichia Coli Bacteria Using Artificial Intelligence (AI)

Abstract

Detection and counting of Escherichia coli (E. coli) bacteria is an important indicator in assessing water quality and food safety, particularly in the field of environmental health. Conventional methods still require a long time, skilled personnel, and have the potential to cause subjectivity in microscopic observations. This study aims to design and build a digital microscope system equipped with an automatic bacterial counting system based on Artificial Intelligence (AI). The use of this technology is expected to accelerate microbiological analysis, improve detection accuracy, and reduce subjectivity in the manual counting process. The study was conducted using an experimental approach with the System Development Life Cycle (SDLC) method. The dataset consists of 3,011 bacterial images divided into 74% training data, 13% validation data, and 13% test data. The object detection model uses YOLOv11 integrated with Roboflow for annotation and dataset management. Test results show that the model achieved a detection accuracy of 94.1% on the test data, indicating good performance in identifying and counting E. coli colonies. The system is also equipped with a Streamlit-based interface to facilitate users in visualizing detection results in real-time. Thus, the design of this AI-based digital microscope can be an effective and efficient solution to accelerate and improve the accuracy of microbiological analysis, especially in the detection of E. coli bacteria in the fields of environmental health and food safety.