Abstract
Plant leaf disease identification and evaluation in a timely and accurate manner is crucial for efficient farming operations for crop yield optimization. Employing the most recent advances in technology, specifically the combination of deep learning and the Internet of Things (IoT), this paper offers an efficient approach to identifying plant diseases. We propose a transfer learning-based deep learning classification model which makes use of pre-trained models including Convolutional Neural Networks (CNN), AlexNet, Residual Networks (ResNet), InceptionV3, and Visual Geometry Group-16 (VGG-16). To provide wider accessibility, high-resolution images of tomato plant leaves displaying disease symptoms are gathered from a dataset and saved in cloud storage using Internet of Things devices. Following the image extraction from the cloud, images are preprocessed using data argumentation, normalization, color space conversion, background removal, and noise removal. Different tomato plant disease classes are classified using the pre- trained models CNN, AlexNet, ResNet, InceptionV3, and VGG-16. The deep learning models’ accuracy is increased using the transfer learning technique, which also reduces the workout duration. The VGG-16 model outperforms other models in the experiment, recognizing plant illnesses with an astounding accuracy of 93.7% on average, proving the efficacy of the suggested approach. This new approach may revolutionize the diagnosis of diseases affecting tomato plants and promote environmentally friendly agricultural practices.
Keywords: Agricultural IoT, Crop Health, Deep Learning Machine Vision, Disease Prediction, Pre-trained Models.