A convolutional neural network approach for detection of E. coli bacteria in water

Abstract

The detection of Escherichia coli bacteria is essential to prevent health diseases. According to the laboratory-based methods, 12–48 h is required to detect bacteria in water. The drawback of depending on laboratory-based methods for the detection of E. coli bacteria can be prone to human errors. Hence, the bacterial detection process must be automated to reduce error. We implement an automated E. coli bacteria detection process using convolutional neural network (CNN) to address this issue. We have also proposed a mobile application for the rapid detection of E. coli bacteria in water that uses CNN. The developed CNN model achieved an accuracy of 96% and an error (loss) of 0.10, predicting each sample in only 458ms. The performance of the model was validated using the F-score, precision, sensitivity, and accuracy statistical measures, which shows that the model is reliable and effective in detecting E. coli. The study generates a methodology for predicting E. coli bacteria in water, which can be used to predict hotspots in terms of continuous exposure to water contamination.