Convolutional Neural Network (CNN) Based Identification of Crop Diseases Using ResNet-50

Agriculture is vital to the success of India's economy. Loss of crops and future disease propagation can be avoided if plant diseases are detected early. For the sake of human health worldwide, early identification of plant diseases is essential. In the past, a pathologist would have to physically inspect each plant to make a diagnosis. However, due to lower precision and limited availability of human resources, manual examination for agricultural diseases is restricted. Automated methods for spotting and labelling plant diseases are in high demand as a solution to this kind of problem. The recent achievements of DL in computer vision for plant disease categorization have served as inspiration for several publications. Unfortunately, the deficit translucence in these end-to-end deep classifiers can limit their practical application. We presented convolutional neural networks using the Resnet50 network to address the issues mentioned earlier. In this study, we use DL for the problem of disease classification in plants. Due to its immense success in image-based classification, the convolutional neural network (CNN) model is used. We can look at ResNet50 with Adam Optimizer to improve the system's accuracy. In this case, crop leaf images are used as input, and the system will “stumble on” the presence of disorder after processing the image. If a disease is found, the results will specify its kind and may suggest solutions like insecticides or other chemicals. The dataset is used to train the CNN model and previously trained models like VGG, ResNet, and DenseNet. The Resnet50 Network model is the most accurate of these. The results demonstrate that the proposed model outperformed state-of-the-art approaches by producing a 99.02% accuracy rate.