Disclosing the Potential of Deep Learning in Cassava Leaf Disease Analysis by using CNN and Neural Networks Approach

The identification and classification of cassava infections are paramount due to their detrimental impact on agricultural productivity. This study conducts a comparative analysis to assess the effectiveness of various deep learning techniques in classifying cassava leaf diseases. Transformer-Embedded ResNet, EfficientNetV2 with visual attention, and a mobile-based model are explored to address the challenges posed by an imbalanced dataset. Leveraging deep Convolutional Neural Networks (CNNs) and attention mechanisms, these models exhibit enhanced precision and effectiveness. Evaluations measure accuracy, precision, recall, and F1-score, considering feasibility and computational complexity. Results indicate that the proposed models effectively control cassava infections compared to current methods. This study underscores the precision and efficacy of plant disease identification using deep learning techniques, presenting modern methodologies for thorough evaluation. With agriculture supporting a significant portion of the world's population, AI-powered automation offers solutions to production challenges. AI enhances agricultural precision, monitors crop health, identifies diseases, and forecasts weather conditions, particularly advantageous in addressing manpower shortages. Automated detection is crucial for mitigating agricultural disease risks and ensuring food security. Focusing on the examination of CNNs and neural networks, this study utilizes deep learning techniques to identify cassava leaf diseases using the Kaggle dataset and real-time photographs. Performance evaluation and enhancement are discussed, with a specific focus on the impact of disease on the Thai cassava crop. Experimental trials demonstrate the effectiveness of deep learning in automating cassava disease categorization, particularly in detecting brown streak virus illness, yielding notable F-measure and accuracy. This study advances disease classification and calls for further research in the field.