International Journal of Electrical and Computer Engineering (IJECE) Vol. 15, No. 1, February 2025, pp. 550~558 ISSN: 2088-8708, DOI: 10.11591/ijece.v15i1.pp550-558  550 Journal homepage: http://ijece.iaescore.com Towards optimal fillet portioning: a computer vision system for determining the fish fillet volume Chanh-Nghiem Nguyen 1,2 , Ngoc-Tan Vo 1,2 , Ngoc-Thanh Nguyen 1,2 , Nhut-Thanh Tran 2 , Chi-Ngon Nguyen 2 1 Automation Laboratory, Can Tho University, Can Tho, Vietnam 2 Faculty of Automation Engineering, Can Tho University, Can Tho, Vietnam Article Info ABSTRACT Article history: Received Apr 11, 2024 Revised Aug 11, 2024 Accepted Aug 20, 2024 Portioning large fish fillets for packaging is usually performed manually by skilled workers. Automating this process and obtaining packaged products with attractive shapes and affordable weights will be beneficial for promoting purchase decisions. Towards developing an automated fish fillet portioning system, this study investigated a computer vision approach for determining the fillet volume. A belt conveyor would transport a fish fillet to the image capture booth, where its cross-section areas would be calculated for volume determination. The developed system could be operated with a conveyor speed ranging from 7.5 to 30.6 mm/s. The system performance was evaluated at a conveyor speed of 7.5 mm/s using small catfish fillets from 142.2 to 225.4 cm 3 . A mean percent error of 9.2% was observed, and the smallest percent error of 3.8% was obtained with a 225.4 cm 3 fillet. With minor measurement errors obtained for larger fillets, the proposed computer vision system showed great potential for developing a cost-effective automated system for customized fish fillet partitioning to accelerate purchase decisions and also for quality control and classification of the fish fillets. Keywords: Catfish Computer vision Fillet portioning Laser light Vision-based system Volume measurement This is an open access article under the CC BY-SA license. Corresponding Author: Chanh-Nghiem Nguyen Faculty of Automation Engineering, Can Tho University Campus II, 3/2 Street, Ninh Kieu District, Can Tho City, Vietnam Email: ncnghiem@ctu.edu.vn 1. INTRODUCTION Fish are among the healthiest foods and are rich in essential nutrients, including high-quality proteins, various fatty acids, minerals, and vitamins [1]. Fish consumption was also linked to health benefits; for example, a reduction in coronary heart disease and stroke was reported with an increased fish intake [2], [3]. As a result, the consumption of fish or seafood in general has been a part of national dietary recommendations in many countries, particularly in most countries in Europe [4]. However, the daily or weekly recommended serving differed in many national guidelines, which could be a fixed amount or within a recommended range [4], [5]. Fish products sold at local markets are available in different presentation formats, such as whole fish, fish steaks, fish fillets, round cuts, and ready-to-cook products. Among these formats, fillets are preferred over alternatives [6], [7], probably due to their ease of preparation [7], [8]. It was reported that consumers were generally willing to pay a premium price for fish fillets [7], [9]. Thus, large fish fillets should be sliced into smaller volumes that fall within the recommended serving ranges. This “good-for- health” information would encourage fish consumption because consumers would consent to paying higher premiums with health claims, as reported in [7].