Cassava Disease Classification Using Squeezenet CNN Technique
Darios B. Alado · 2024
Agriculture plays a pivotal role in the economic development of South Asian and African nations. However, adverse climate changes and disease invasions threaten crop productivity annually. Early identification and management of crop diseases are crucial for mitigating these effects. Cassava, a significant crop in the Philippines and other regions, faces challenges from various diseases, necessitating advanced diagnostic methods. This study uses deep learning techniques, specifically Convolutional Neural Networks (CNNs), to identify cassava leaf diseases. Researchers employed the Squeezenet architecture due to its efficiency and suitability for deployment on mobile devices. Our model classified cassava leaf status into four categories: Healthy, CBB (Cassava Bacterial Blight), CPD (Cassava Phytoplasma Disease), and BLS (Bacterial Leaf Streak). Using a robust dataset, the model achieved a high overall accuracy of 92%. The confusion matrix indicated strong performance in identifying diseased leaves, though there were some misclassifications in the Healthy category. These findings demonstrate the potential of deep learning techniques in enhancing crop disease management. They offer a promising tool for farmers in developing nations to protect their crops and improve agricultural productivity.