ILKOM Jurnal Ilmiah Vol. 14, No. 1, April 2022, pp. 80-90 Research Article Open Access (CC–BY-SA) Abstract Microorganisms such as bacteria are the main cause of various infectious diseases such as cholera, botulism, gonorrhea, Lyme disease, sore throat, tuberculosis and so on. Therefore, the identification and classification of bacteria are very important in the medical field to help doctors diagnose diseases suffered by patients. However, manual identification and classification of bacteria takes a long time and a professional individual. With the help of artificial intelligence, we can effectively and efficiently classify bacteria and save a lot of time and human labor. In this study, a system was designed to classify bacteria from microscopic image samples. This system employed deep learning with the transfer learning method. Inception V3 architecture was modified and retained using 108 image samples labeled with five types of bacteria, namely Acinetobacter baumanii, Escherichia coli, Neisseria gonorrhoeae, Propionibacterium acnes and Veionella. The data were then divided into training and validation using the k-fold cross validation method. Furthermore, the features that have been extracted by the model were trained with the configuration of minibatchsize 5, maxepoch 5, initiallearnrate 0.0001, and validation frequency 3. The model was tested with data validation by conducting ten experiments and obtaining an average accuracy value of 94.42%. Accredited 2 nd by RISTEKBRIN No. 200/M/KPT/2020; E-ISSN 2548-7779 | P-ISSN 2087-1716 Multi Classification of Bacterial Microscopic Images using Inception V3 Ingrid Nurtanio a,1,* ; Anugrayani Bustamin a,2 ; Christoforus Yohannes a,3 ; Alif Tri Handoyo a,4 a Universitas Hasanuddin, Jl Perintis Kemerdekaan KM 10, 90245, Makassar, Indonesia 1 ingrid@unhas.ac.id; 2 anugrayani@unhas.ac.id; 3 christoforus@unhas.ac.id; 4 alif.trihandoyo15@gmail.com * Corresponding author Article history: Received March 1, 2022; Revised April 4, 2022; Accepted April 4, 2022; Available online April 30, 2022 Keywords: Bacterial Classification; Deep Learning; Inception V3; Transfer Learning; Image Processing Introduction According to the Food and Agriculture Organization (FAO), the number of victims who die from bacterial infections reaches up to 700,000 people every year. Recognition of bacterial genera and species is necessary because knowledge of the biology of microorganisms is significant in medical field, veterinary medicine, biochemistry, the food industry, and agriculture. Although most microorganisms have a positive impact on various areas of life, they can cause many diseases, including infectious diseases [1] [2]. Biologists identify and classify different types of bacteria with different biochemicals and forms. They used different bacterial attributes for classification, for example, the shape of bacterial cells (spiral, cylindrical and spherical). The size and structure of the colonies formed by bacteria are examined to distinguish bacterial species. Cells of several types of bacteria have different sizes and structures depending on environmental conditions. Several species of bacteria have very similar shapes. Although each bacterial species has its own characteristics, the biochemical reactions carried out by bacteria and their metabolic activities together help classify the species. However, the classification of bacterial species is not an easy task even for an experienced specialist [3] [4]. Song, et al. have proposed analysis for Bacterial Vaginosis (BV) images from a microscope. They proposed an automatic method to diagnose BV with several stages, i.e., segmentation, splitting, and classification of overlapping bacteria image cases. Following that, the implementation of the Nugent score criterion was used in their experiment demonstration and achieved high accuracy with computing efficiency [5]. In general, microbiological image analysis using traditional laboratory methods has bacteria recognition errors and requires different experience and long processing time. Therefore, the automatic classification technique of bacterial images is more valuable than traditional visual observations for biologists because of its accurate classification, low cost, and fast diagnosis. Previous research related to bacterial classification was carried out in 2018 by Basma, et al. classifying ten types of bacteria using the Bag of words feature extraction and Support Vector Machine (SVM) method with 97% accuracy [6]. https://doi.org/10.33096/ilkom.v14i1.1120.80-90 80