Early Chilli Disease Detection Based on Leaves Images Using Convolution Neural Network (CNN)

Abstract

Chilli is one of the important crops in Malaysia, as it is a main spice used in many local dishes and contributes to the country's economic growth. However, chili plants are prone to diseases, which require early prevention to minimise yield loss. This study presents a CNN-based framework for the early detection of chilli plant diseases based on visible leaf symptoms. Early diagnosis is critical to prevent severe crop loss and reduce the misuse of pesticides. Two deep learning architectures—InceptionV3 and AlexNet were evaluated using a publicly available Kaggle dataset comprising five categories: healthy, bacterial, fungal, viral, and pest-infected leaves. Data augmentation techniques such as flipping, rotation, and scaling were applied to improve generalization. Both models were trained using transfer learning, with modified final layers to adapt to the specific classification task. InceptionV3 achieved the highest performance with an accuracy of 99.25%, slightly outperforming AlexNet’s 98.23%. Evaluation metrics including precision, recall, and F1-score further confirmed InceptionV3’s superior capability in extracting relevant features for accurate classification. This system has potential applications in mobile-based diagnosis tools and autonomous field robots. Future work will focus on deploying lightweight models for real-time detection and integrating pesticide recommendation systems for infected plants.