Fruit recognition with its variety classification is a promising area of research. This research is useful for monitoring and categorizing the fruits according to their kind with the assurance of fast production chain. In this research, we establish a new high-quality dataset of images containing the five most popular oval-shaped fruits with their varieties. Recent work in deep neural networks has led to the development of many new applications related to precision agriculture, including fruit recognition. This paper proposes a classification model for 40 kinds of Indian fruits by support vector machine (SVM) classifier using deep features extracted from the fully connected layer of the convolutional neural network (CNN) model. Also, another approach based on transfer learning is proposed for recognition of Indian Fruits. The experiments are carried out in six most powerful deep learning architectures such as AlexNet, GoogleNet, ResNet-50, ResNet-18, VGGNet-16 and VGGNet-19. So, the six deep learning architectures are evaluated in two approaches, which makes 12 classification model in total. The performance of each classification model is assessed in terms of accuracy, sensitivity, specificity, precision, false positive rate (FPR), F1 score, Mathew correlation (MCC) and Kappa. The evaluation results show that the SVM classifier using deep learning feature provides better results than their transfer learning counterparts. The deep learning feature of VGG16 and SVM results in 100% in terms of accuracy, sensitivity, specificity, precision, F1 score and MCC at its highest level.
Behera, Santi Kumari; Rath, Amiya Kumar; and Sethy, Prabira Kumar
"Fruit Recognition using Support Vector Machine based on Deep Features,"
Karbala International Journal of Modern Science: Vol. 6
, Article 16.
Available at: https://doi.org/10.33640/2405-609X.1675
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.