Abstract

Diseases in plants are one of the biggest threats to food safety. A number of the conditions in plants are infectious diseases thatmay spread throughout the whole field, thus affecting most of the yields. Therefore, early diagnosis of diseases in plants isessential. The proposed system has developed a Deep learning-based (DL) disease classification system by using Xceptionarchitecture, which can correctly identify the diseases in plants when a picture of the infected area is given as input. Apart fromXception architecture, we also compare Inception ResNet V2, RNN, ResNet50, Efficientnet hybrid using Inception andXception to determine which model performs best and might provide the very best accuracy. For the identification of plant leafdiseases, Xception is proposed in this research. These suggested models are trained using the Plant Dataset, which contains500 photos of healthy and ill leaves in 38 different classes. Up to 95.81% accuracy in illness classification has been achievedby the suggested architecture, and various observations have been made with different algorithms such as ResNetV2, RNN,ResNet50, Xception & EfficientNet. The experiment's results are comparable to those of other procedures that have beenpublished.Training the computer on the obtained data would be done by using Deep learning techniques. On the newly-trainedmachine, the obtained data would be tested. If the leaves look to be contaminated, the system can tell. Another useful methodfor recommending crop varieties to farmers is described in this study. Input from the user includes the location and type of soil.Machine learning algorithms can be used to select the most lucrative crop list, or a crop recommendation can be projected fora crop that the user specifies. Crop recommendation estimates are made possible by the application of machine learningtechniques such as Random Forest (RF), Decision Tree (DT), KNN, Naive Bayes, and voting classification. Random Forestgave the best results in terms of accuracy, with a 95% success rate.Keywords: ResNet, Plant disease identification, Deep Learning, Random Forest