A Deep Learning Model for Multiple Apple Foliar Diseases Identification

Abstract

Apple orchards are facing threats of pathogens and insects, which costs millions of dollars lost every year in the United States. The current apple disease diagnose process is based on naked eyes. The time-consuming process requires apple farmers to have professional knowledge about various apple diseases. Incorrect diagnosis will cause chemical abuse, environmental pollution, and financial loss. Because of the involution of deep learning, fast and efficient plant diseases detection powered by a deep learning model is possible. In the past few years, many high accurate deep learning models achieved high performance on apple foliar diseases classification problems. But these models have a limitation that is they only can detect a single apple disease. The paper focuses on implementing a deep learning model for multiple apple foliar diseases identification problems according to the Plant Pathology 2021 competition on Kaggle. The model achieved high accuracy on both singly and multiply apple foliar diseases identification problems.