Image Processing Identification for Sapodilla Using Convolution Neural Network (CNN) and Transfer Learning Techniques
Ali Khazalah, Boppana Prasanthi, Dheniesh Thomas, Nishathinee Vello, Suhanya Jayaprakasam, Putra Sumari, Laith Abualigah, Absalom E Ezugwu, Essam Said Hanandeh, Nima Khodadadi
17 November 2022
Image identification is a useful tool for classifying and organizing fruits in agribusiness. This study aims to use deep learning to construct a design for Sapodilla identification and classification. Sapodilla comes in a various of varieties from throughout the world. Sapodilla can come in different sizes, form, and taste depending on species and kind. The goal is to create a system which uses convolutional neural networks and transfer learning to extract the feature and determine the type of Sapodilla. The system can sort the type of Sapodilla. This research uses a dataset including over 1000 pictures to demonstrate four different types of Sapodilla classification approaches. This assignment was completed using Convolutional Neural Network (CNN) algorithms, a deep learning technology widely utilised in image classification. Deep learning-based classifiers have recently allowed to distinguish Sapodilla from various images. Furthermore, we utilized different versions of hidden layer and epochs for various outcomes to improve predictive performance. We investigated transfer learning approaches in the classification of Sapodilla in the suggested study. The suggested CNN model improves transfer learning techniques and state-of-the-art approaches in terms of results.