Integrating Computer Science into Biology: A Deep Learning Model for Bacterial Shape Classification in STEM Education

There is a rising demand to close the gap between biology and computer science in high school STEM education. Biology is often taught with a textbook and a microscope. To help students recognize the spiral, bacilli, and cocci shapes of bacteria, this study introduces an interdisciplinary approach by incorporating a deep learning image classification model into the biology classroom. An 86% model accuracy was achieved by using a pre-trained Xception model that was refined on a dataset of bacterial images. The model was employed as a teaching aid to improve students’ understanding of bacterial morphology by directly comparing the classifications of the trained model and humans. A lesson plan lets students label bacteria manually, analyze the model’s predictions, and discuss the biological structure and visual patterns. This study introduces interdisciplinary STEM projects by demonstrating how machine learning can enhance biology instruction and introduce students to AI applications in science.