7th ISF

COMP026 - Image Classification by Using Deep Learning to Identify the Presence of Pseudomonas aeruginosa Bacteria in Soil and Water to Prevent Disease in Plants and Animals

The bacteria Pseudomonas aeruginosa is known to cause diseases in plants, animals, and humans, significantly impacting agricultural and environmental ecosystems. In response, this project applies deep learning techniques, particularly Convolutional Neural Networks (CNN) and the VGG-16 model, to detect the presence of P. aeruginosa in soil and water samples. The study encompassed several key steps, commencing with the collection and cultivation of P. aeruginosa samples from diverse sources. These samples were utilized to capture colony images, subsequently divided into training and testing datasets to develop, and evaluate the deep learning model’s accuracy. The dataset included images of P. aeruginosa colonies both in isolation and within mixed bacterial colonies, enabling the model to recognize P. aeruginosa even in complex sample mixtures. The primary objective was to develop a deep learning model capable of achieving a prediction accuracy exceeding 90%. The results revealed that the VGG-16 model achieved a maximum classification accuracy of 76.47% using a dataset comprising over 100 instances of both P. aeruginosa and non-P. aeruginosa samples: Bacillus subtilis, Escherichia coli, and mixed bacteria, divided into an 80:20 train-test split.

Show More