Design of efficient classification model for Paramecium and Hydra microorganisms

This paper presents a novel approach to microorganism image classification, concentrating on Paramecium and Hydra bacteria. Due to their intricate features and minute variations, these microorganisms—which are crucial to biological research and environmental studies—present particular difficulties in picture classification. There is a noticeable class imbalance in the dataset in this study, which includes 152 photos of Paramecium and 76 images of Hydra. Generative Adversarial Networks (GANs) are employed for data augmentation, generating synthetic examples to rebalance the dataset. This method increases the quantity of the dataset and adds a variety of examples, which strengthens the model’s capacity for generalization. Transfer learning is explored using Inceptionv3 and ResNet50, along with machine learning techniques such as Adaboost and XGBoost. Convolutional Neural Networks extract discriminative features, and GANs enhance classification performance. The proposed approach achieves an impressive accuracy and F1-score of 96.49% and 96.77%, respectively, in accurately distinguishing between hydra and paramecium bacteria. This research not only demonstrates a high degree of accuracy in classifying microorganism images but also contributes significantly to the field by proposing a robust solution to the challenge of class imbalance.