Multi-Phase Plant Leaf Disease Detection: Leveraging Feature Engineering, Transfer Learning, and Explainable AI
Abstract
— Early and accurate detection of plant diseases is crucial for agricultural productivity. This paper presents a novel multi-phase research approach for plant leaf disease detection. We progressively explore the potential of traditional machine learning, deep learning with transfer learning, and explainable AI techniques. In the first phase, we utilize feature engineering and a machine learning neural network model to achieve 97% accuracy on an apple leaf disease dataset. The second phase leverages a pre-trained ResNet-50 model with a novel 9-layer ResNet block, achieving 99.01% accuracy on the PlantVillage dataset. Finally, the third phase employs a variation with Explainable AI integration, reaching 99.09% accuracy on the apple leaf dataset and providing valuable insights into the model's decision-making process. This research demonstrates the effectiveness of a progressive approach, highlighting the potential of transfer learning and explainability in deep learning for plant disease detection
Keywords
Plant Disease Detection, Feature Engineering, Machine Learning, Deep Learning, Transfer Learning, ResNet, Explainable AI