Diseases Detection from Apple Leaf using Deep Transfer Learning Approach

Abstract

Apples are one of the topmost eaten fruits around the globe and have a massive global production value. Although fertilizers and pesticides have improved the quality and quantity of apple production, certain diseases still need to be caught beforehand. This disease can cause devastation to fruit production. They also have chances to become life-threatening for the trees as well. Advances in imaging technologies are growing, and it is necessary to dedicate these state-of-the-art technologies into the agricultural division. This research presents the utilization of a deep learning algorithm for the detection of three major apple leaf diseases. The analysis utilizes image processing techniques such as K-mean clustering for segmentation. GoogleNet architecture has been used for the classification of the processed data set. A systematic performance of the proposed approach was also determined with the following accuracy of 98.54%, precision of 98.50%, recall of 98.50%, and F1-score of 98.50. This study will equip the farmers with a proper and effective method for early apple leaf disease detection to be able to act in good time to ensure the crops' safety, improvement in productivity, and sustainable agriculture.