Comparative Study Of Deep Learning Algorithms For Disease And Pest Detection In Rice Crops
Syeda Airas Burhan, Sidra Minhas, Amara Tariq, Muhammad Nabeel Hassan
Abstract
Accurate and timely detection of rice crop diseases like Leaf Blast and Brown Spot, as well as pests like Hispa, can help minimize crop losses and increase the yield obtained. Since Pakistan is an agricultural country, such researches are imperative to its economic growth. This research is focused on a comparative study between the performances of five different Deep Learning Models i.e. Vgg16, Vgg19, ResNet50, ResNet50V2, and ResNet101V2 on both artificial data as well as on images collected from the rice fields in Gujranwala, Pakistan. The artificial data set has been classified into four classes Hispa, Healthy, Brown Spot, and Leaf Blast; whereas binary classification of Healthy Vs. Unhealthy has been performed on the data set collected from the fields. All images have been pre-processed by removing backgrounds and shadows before being passed through the models. On the artificial data set, the ResNet50 model performed the best with an accuracy of 75.0real data set, the ResNet101V2 was the best performing model with an accuracy of 86.799.