Study and Analysis of Plant Disease Identification Models

Agriculture is crucial to modern life in the broader economic landscape, with crop diseases causing significant economic losses. Diagnosing plant diseases accurately remains challenging, prompting research into efficient and rapid solutions such as Deep Learning (DL). DL, widely applied in various fields, including agriculture, offers the potential for high accuracy and efficiency in disease diagnosis. In this paper, the analysis of model accuracy for the ResNet50V2, VGG16, DenseNet121 and MobileNetV2 models has been done for different batch sizes. The obtained observations of the analysis indicate that the accuracy is improved from 99.1% to 99.45% by changing the batch size from 32 to 64. The proposed change for batch size utilizes the newer and capable GPUs more efficiently, thus providing increase in accuracy.