Deep Learning in Agritech: Exploring Techniques and Architectures for Plant Disease Detection

Abstract

Disease detection in agriculture is crucial for maintaining crop health and productivity. Traditional methods of detection rely on manual observation and expert knowledge, which can be time-consuming and subjective. In recent years, deep learning techniques have emerged as a promising approach for automated disease detection. This paper investigates the performance of three state-of-the-art deep learning models, ResNet34, MobileNetv2, And Yolov8, in detecting and categorizing leaf diseases among corns. The study revealed that MobileNet2 demonstrates computational effectiveness in classifying corn leaf diseases, with a low average loss and the ability to discriminate between specific illnesses. Yolov8 is identified as a strong candidate for real-world applications due to its consistently high precision (95.73%), recall (94.81%), accuracy (96.18%), and F1 score (95.22%) across all disease categories. It excels in distinguishing between Blight and Common Rust and accurately recognizing healthy leaves. ResNet34 and MobileNetv2 also show competitive results, but Yolov8 performs better overall, particularly in terms of precision and recall. However, the choice of the optimal model depends on the specific application's demands, computational effectiveness, and the balance between recall and precision. Further investigation and refinement endeavors are recommended to augment the potential of these models and address any remaining obstacles in disease identification.