This paper presents an artificial intelligent system for detection of apple diseases using Support Vector Machine (SVM) and Cockroach Swarm Optimization (CSO). This paper faces a challenge of content ambiguity. This challenge is mitigated by combining texture classification based on initial k-means clustering. The proposed system is able to extract useful features. There are 4 classes: scab disease, rot disease, blotch disease, and normal. The dataset consists of 320 apple images in total divided into 80 images for each class. SVM is actually a binary classifier. The paper shows how to extend it for multi-class classification. The performance of SVM strongly depends on proper choice of its parameters. CSO meta-heuristic performs fine-tunes for SVM parameters including nonlinear hyper planes, penalty parameter of error term, and degree which is used to find the hyper plane for splitting the data. Results show 94.65% accuracy in detecting the normal apple.
CITATION STYLE
El-dosuky, M. A., Oliva, D., & Hassanien, A. E. (2020). An Artificial Intelligence System for Apple Fruit Disease Classification Based on Support Vector Machine and Cockroach Swarm Optimization. In Advances in Intelligent Systems and Computing (Vol. 1153 AISC, pp. 137–147). Springer. https://doi.org/10.1007/978-3-030-44289-7_14
Mendeley helps you to discover research relevant for your work.