Use of YOLOv8 architecture for identification of lactic acid-producing strains

This paper explores the use of the YOLOv8 architecture for the identification of lactic acid-producing strains, which is of significant importance for the food and pharmaceutical industries. We developed a deep neural network method for the rapid and accurate classification of microorganisms based on their visual characteristics. A dataset of images of various lactic acid strains was collected and annotated. To enhance model accuracy, data augmentation techniques and pre-training on large image datasets were employed. Experiments showed that the mean Average Precision (mAP) reached 89%, while the recall was 84%, confirming the high effectiveness of the proposed approach. This study demonstrates the potential of modern computer vision technologies for automating the identification of microorganisms, which can significantly accelerate and simplify laboratory and production processes. The results obtained open new prospects for further research in microbiology and biotechnology.