Enhanced Support Vector Machine Based on Grey Wolf Optimizer for Fruits Image Classification using MPEG-7 Color and Texture Features Fusion

Abstract

Fruit classification from images plays a pivotal role in diverse domains. Despite numerous efforts to tackle this challenge, it remains complex due to the diversity of fruit and applications. This study presents an enhanced support vector machine (SVM) based on grey wolf optimizer (GWO) for fruit image classification. GWO is used to optimize the hyperparameters of SVM and low variance feature selection threshold. The utilization of MPEG-7 visual descriptors negates the need for segmentation. The results showcase exceptional classification accuracy across Ubaya-IFDS3000, Ubaya-IFDS5000, and Supermarket produce datasets, with standout features achieving up to 99.21%, 98,28%, and 99.85% accuracies, respectively. Notably, the proposed method consistently outperforms SVM optimized with the other optimization algorithms. Further, it excels in classification accuracy when compared to previous state-of-the-art methods. This study emphasizes the importance of hyperparameter optimization using GWO and its effectiveness in fruit image classification.