AUTOMATIC BANANA LEAF DISEASE CLASSIFICATION USING EFFICIENTNET

Plant leaf diseases pose a serious challenge to global food security by significantly reducing both the yield and quality of agricultural produce. Conventional disease identification methods rely heavily on manual inspection of plant leaves, which is time-consuming, subjective, and prone to human error. The development of rapid, accurate, and automated disease detection systems using computer vision and artificial intelligence (AI) offers a promising alternative to overcome these limitations. Banana cultivation plays a vital role in commercial agriculture, particularly in Asian and African regions. While banana production is influenced by environmental factors and natural disasters, plant diseases remain a persistent threat that adversely affects productivity and crop quality. Over the past decade, image processing and machine learning approaches have been widely explored for plant disease identification and classification.In this study, a deep learning–based framework is proposed for the classification of banana leaf diseases. The system follows a structured workflow comprising image acquisition, preprocessing, feature extraction, and classification. A custom dataset was developed using banana leaf images collected directly from agricultural fields, consisting of two categories: healthy leaves and leaves affected by Sigatoka disease. An EfficientNet-based deep learning model was employed to classify the leaf images into their respective classes. The proposed model demonstrated strong performance, achieving a training accuracy exceeding 99.59% and a testing accuracy of over 98.12%. These results highlight the effectiveness of deep learning models in accurately identifying banana leaf diseases and support their potential application in real-world agricultural settings.