Quality recognition method of oyster based on U-net and random forest

• Published in: Journal of food composition and analysis Vol. 125; p. 105746
• Main Authors: Zhao, Feng, Hao, Jinyu, Zhang, Huanjia, Yu, Xiaoning, Yan, Zhenzhen, Wu, Fucun
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.01.2024
• Subjects: Consumer preference; Oyster; Quality recognition; Random forest; U-Net; Quality recognition; Random forest; Oyster; U-Net; Consumer preference
• ISSN: 0889-1575; 1096-0481
• DOI: 10.1016/j.jfca.2023.105746

Oysters are one of the most important cultivated marine resources globally. The shape of oysters is an essential reference criterion for consumers to judge the quality of oysters. In order to recognize oyster’s shape, the U-Net model and random forest are combined to compose a creative strategy. To be more specific, the U-Net neural network model is firstly developed to segment the image and obtain the contours of oysters, and the shape features of oysters are extracted. Then, a random forest model with shape feature parameters depending on customer preference is created to identify oyster quality. The results indicate that the intersection-over-union of segmentation outcomes achieved by U-Net reaches 99.06%, surpassing the 93.50% obtained by traditional methods. The accuracy of the classification strategy based on the shape features parameters of consumer preference is 94.18%, which further proves the effectiveness of the proposed strategy. This study might provide valuable data and guidelines to oyster product classification based on shell shape within market contexts. ●A new quality identification method of oyster based on machine vision is proposed.●Extraction of oyster contours by U-Net to raise the accuracy of shape parameters.●A strategy of oyster quality classification based on customer preferences.●Random forest model relying on customer preference identify oyster quality.