Performance Evaluation of Pre-trained Convolutional Neural Network for Milkfish Freshness Classification

Prasetyo, Eko and Purbaningtyas, Rani and Adityo, R Dimas (2021) Performance Evaluation of Pre-trained Convolutional Neural Network for Milkfish Freshness Classification. Information Technology International Seminar (ITIS). pp. 30-34. ISSN 978-1-7281-7726-7

	Text itis2020 (cek plagiasi).pdf Download (1MB)
	Text Bukti Corr Author ITIS 2020.pdf Download (776kB)
	Text Bagian Depan ITIS 2020.pdf Download (270kB)
	Text 2020-Performance Evaluation of Pre-trained Convolutional Neural Network for Milkfish Freshness Classification.pdf Download (268kB)

Abstract

Milkfish are the top five fish of aquaculture products in Indonesia with high sales in traditional markets. Hence, the Indonesian people should recognize the freshness of the fish in the traditional market. An automated system to recognize the freshness of milkfish based on the eye using Convolutional Neural Network (CNN) deep learning requires vast image data in training sessions. For our small dataset, we performed transfer learning with fine-tuning pre-trained CNNs. In this study, we evaluate several pre-trained CNN models to classify milkfish eye freshness. The dataset consists of 234 milkfish eye images and three freshness class. The experiments and analysis results show that NasNet Mobile and Densenet 121 outperform state-of-the-art with the best performance on training, validation, and testing data.

Item Type:	Article
Uncontrolled Keywords:	Convolutional Neural Network, milkfish eye, freshness, classification, transfer learning
Subjects:	Technology
Divisions:	Faculty of Engineering
Depositing User:	Perpus Ubhara Surabaya
Date Deposited:	25 Jun 2024 03:10
Last Modified:	25 Jun 2024 03:20
URI:	http://eprints.ubhara.ac.id/id/eprint/2554

Actions (login required)

View Item