A Scheme Based on Deep Learning for Fruit Classification

Ali Orangzeb Panhwar; Anwar Ali Sathio; Nadeem Manzoor Shah; Sumaira Memon

doi:10.22581/muet1982.2742

Ali Orangzeb Panhwar Faculty of Computing and Engineering Sciences, SZABIST University Gharo, Pakistan
Anwar Ali Sathio Department of Computer Science and Information Technology, Benazir Bhutto Shaheed University, Karachi, Sindh, Pakistan
Nadeem Manzoor Shah e Department of Civil Engineering, Mehran University of Engineering & Technology, Jamshoro Sindh, Pakistan
Sumaira Memon Dr. AHS Bukhari, Faculty of Engineering and Technology University of Sindh, Jamshoro, Pakistan

DOI: https://doi.org/10.22581/muet1982.2742

Keywords: CNN, VGG16, VGG19, Fruit, Grading, Deep Learning

Abstract

Grading and classifying fruits are critical due to automated machine learning systems. In computer vision, different fruits have large complexity and similarity to identify the fruit types. In this study, we developed an efficient and reliable fruit grading system. It is very difficult to classify fruits from images with established conventional approaches. We used a Convolutional Neural Network (CNN) methodology involving comparing a custom-built CNN and the VGG pre-trained models. In the research results, the VGG model accuracy is of 99.98 percent. This research proved the effectiveness of the deep model in the challenges of fruit classification and set a foundation for its application in automated grading systems.