IMAGE PROCESSING AND DEEP LEARNING BASED CLASSIFICATION OF COFFEE LEAF DISEASE

Abstract

Coffee leaf diseases are a major threat to coffee production in Ethiopia and worldwide. Early detection and treatment of diseases are essential to prevent crop losses. Convolutional neural networks (CNNs) are a powerful machine learning technique that can be used for image classification. In this research report, we explore the use of CNNs for coffee leaf disease identification. We show that CNNs can be used to achieve high accuracy on this task, even with a relatively small dataset. We also show that AlexNet is a good choice for the base architecture of CNNs for coffee leaf disease identification. The approach is based on AlexNet architecture, and it achieved an accuracy of 97.5% on a dataset of 12600 coffee leaf images. Our research has several implications for the use of CNNs for coffee leaf disease identification. First, it suggests that CNNs are a promising new approach for this task. Second, it suggests that AlexNet is a good choice for the base architecture of CNNs for this task. Third, it suggests that the use of larger datasets can further improve the accuracy of CNNs for this task. Our research also has several limitations. First, our dataset was relatively small. This means that the models we trained may not be able to generalize well to new data. Second, we only evaluated our models on a single type of coffee leaf disease. It is possible that the models would not perform as well on other types of coffee leaf diseases. Despite these limitations, our research provides a good foundation for future research on the use of CNNs for coffee leaf disease identification. We believe that CNNs have the potential to revolutionize the way that coffee leaf diseases are identified and managed.