Comparative analysis between Convolutional Neural Network (CNN) and Starfish Optimized with CNN for citrus fruit disease detection and classification

Abstract

Agriculture is a backbone of our country. Many farmers face different challenges in their crop cultivation due to limited water resources, plant diseases, climate change, improper fertilization. This study focuses on citrus fruits. Citrus crops play a vital role in the food production and contribute significantly to the agricultural sector.  Early detection of citrus disease helps prevent the yield reduction and economical losses. This research presents a deep learning approach using Convolutional Neural Network (CNN) for the automatic detection and classification of citrus diseases such as black spot, canker, greening, scab. In this paper, primary dataset was used, containing labeled images of both diseased and healthy citrus fruits. In this proposed system, deep learning approached based on convolutional neural network (CNN) model is used for classification and detection of citrus disease. Starfish optimization algorithm (SFOA) is introduced to tune the hyperparameters in CNN. CNN model achieved a training accuracy of 92.44 % with loss of 0.2704, demonstrating high performance in distinguishing among multiple citrus disease classes. In comparison of CNN, CNN with SFOA achieved 96.55 % with loss0.0739, demonstrating high performance in distinguishing among multiple citrus disease classes. This paper provides an efficient, scalable and reliable tool for early citrus disease diagnosis, which can assist farmers and agricultural experts in real-time monitoring and management of citrus crops.