Multilayer Convolutional Neural Network Based Approach to Detect Apple Foliar Disease

Abstract

Around 38% of land in the world is used for agriculture and the whole world is completely dependent on agriculture. So, that’s why good crop yield is very important to get high agricultural output. A single disease in a plant can lower crop yield. So, to maintain a high agricultural output, we need to detect disease at the early stage so that the agricultural output should be maintained. There are multiple ways to detect plant disease like detecting a plant disease by using the naked eye by hiring an expert, or by using Artificial Intelligence (AI). By using AI, it takes less time to detect plant disease as compared to detecting using the naked eye. Deep Learning (DL), the sub-branch of AI gives an accurate result as compared to the other sub-branches of AI. In DL, Convolutional Neural Network or CovNet is the latest and revolutionary algorithm to perform this task. An apple tree disease detection model, based on Multilayer CNN, is presented in the paper. To train the proposed Multilayer CNN model, the data is collected from FGVC8 dataset from Plant Pathology 2021, a Kaggle Competition which is supported by the “Cornell Initiative for Digital Agriculture Decision Trees, Logistic Regression, and Random Forests are machine learning algorithms that are compared with the performance of the proposed model. This study shows that the proposed model outperforms Machine Learning algorithms with the accuracy of 91%, Precision of 89%, Recall of 85% and F1-Score of 88.34%.