Pepper Leaf Diagnosis Using Deep-Net with Low-Dimensional Image Classification

Pepper leaves provide a range of health advantages, such as anti-inflammatory properties and digestive support. Nonetheless, certain pepper leaves can be susceptible to bacterial infection, which could potentially undermine these health-promoting qualities. This paper introduces a deep learning architecture designed to automatically classify Pepper leaf images as either healthy or infected with bacteria. Specifically, the DeepNet model is fine-tuned using domain-specific Pepper leaf images. Additionally, Principal Component Analysis (PCA) is applied to the DeepNet input images to extract and select the most discriminating features, leading to improvements in classification accuracy, training speed, and memory usage. The experimental results demonstrate the effectiveness of the proposed compact DeepNet in accurately classifying Pepper leaves, achieving an accuracy rate of 91.2% and an F1-score of 91.0%.