Automatic Detection of Escherichia coli Bacteria from Tryptic Soy Agar Image Using Deep Learning Method

Escherichia coli is a normal bacterial flora that lives in the human intestine, is harmless and is part of a healthy digestive tract. However, there are several strains of pathogenic Escherichia coli that can cause infections in the digestive tract, namely diarrhea. Diarrheal disease in Indonesia needs treatment and study because most of the diagnoses are still based on clinical diagnosis. Conventional methods used for the detection of Escherichia coli bacteria include culture methods, biochemical tests, and serological tests. This method has the disadvantage of requiring a long time, a large number of samples, and a relatively high error in reading the results. Therefore, the detection process needs to be done automatically using the Faster R-CNN deep learning method. In this research, we used Faster R-CNN with Inception v2 and ResNet-50 architecture and added augmentation and Image Enhancement to the Tryptic Soy Agar image dataset. The test results show that the addition of Image Enhancement greatly affects model performance and the model that has the best performance and is most appropriate to use is the Faster R-CNN ResNet-50 architecture with the addition of Contrast Stretching and Gaussian Filters to the image dataset. This model has 91% accuracy, 90% precision, 95% recall, and 92% F-1 score.