An Explainable Deep Learning Network With Transformer and Custom CNN for Bean Leaf Disease Classification

The rapid and accurate classification of microorganisms is essential in medical diagnostics, biological research, and environmental monitoring. This project introduces a Flask-based web application that enables the identification and classification of microorganisms using a hybrid deep learning and machine learning model. Several convolutional neural network architectures were trained and evaluated, including CNN, MobileNet, DenseNet, and ResNet. Among these, the most effective model combined ResNet for deep feature extraction with a Random Forest classifier for final prediction. This hybrid approach outperformed individual models in terms of accuracy and robustness. The system allows users to register, log in, and upload microorganism samples through an intuitive web interface. Once uploaded, the system processes the input using the trained ResNet model to extract features, which are then classified using the Random Forest model into one of eight microorganism categories: Amoeba, Euglena, Hydra, Paramecium, Rod Bacteria, Spherical Bacteria, Spiral Bacteria, or Yeast. The architecture is modular, scalable, and designed for practical use in laboratory or academic settings. With real-time predictions and high accuracy, this system provides a reliable tool for efficient microorganism identification and analysis.

Keywords Microorganism classification, ResNet, Random Forest, Flask, Hybrid model, Deep learning, CNN, Medical diagnostics, Biological research, Image classification.