Exploring the Performance of VGG16 and Efficient Net Models for Plant Disease Classification: A Comparative Approach

Crop production and food security are both seriously threatened by plant diseases. Convolutional neural networks (CNNs) have gained popularity as a potential tool for automating the process of identifying plant diseases thanks to the development of deep learning techniques. In this study, we investigate the use of the VGG16 and Efficient Net CNN architectures for plant disease classification and compare their results. The research uses a massive dataset that includes many types of plants and diseases. To prepare the dataset for further analysis, we add annotations to the photos and split the data into a training set, a validation set, and a test set. Using transfer learning and the ImageNet-obtained pre-trained weights, the VGG16 and Efficient Net models are trained on the same dataset. Accuracy is only one of the measures we use to assess the models' efficacy. In terms of plant disease categorization, both VGG16 and Efficient Net show very good accuracy in experiments. However, they perform in notably different ways. With a training accuracy of 96.06% and a Testing accuracy of 95.84%, Efficient Net surpasses VGG16, as shown by our comparison research. The Efficient Net model outperforms VGG16 in terms of accuracy while using many fewer parameters and being much simpler to run. These results suggest that Efficient Net is a better tool for identifying and classifying plant diseases. Finally, this paper presents a complete analysis contrasting the VGG16 model with the Efficient Net model for identifying plant diseases. The findings demonstrate that Efficient Net outperforms other similar methods and that this has practical implications for detecting plant diseases in outside environments.