Classification of Chicken Carcass Breast Blood-Related Defects Using Hyperspectral Imaging Combined with Convolutional Neural Networks

被引:2
作者
Duan, Liukui [1 ]
Bao, Juanfang [1 ]
Yang, Hao [1 ]
Gao, Liuqian [1 ]
Zhang, Xu [1 ]
Li, Shengjie [2 ]
Wang, Huihui [1 ]
机构
[1] Dalian Polytech Univ, Sch Mech Engn & Automat, Dalian 116034, Peoples R China
[2] Dalian Polytech Univ, Sch Food Sci & Technol, Dalian 116034, Peoples R China
基金
中国国家自然科学基金; 国家重点研发计划;
关键词
chicken carcass breast; defect detection; hyperspectral imaging; deep learning; CNN;
D O I
10.3390/foods13233745
中图分类号
TS2 [食品工业];
学科分类号
0832 ;
摘要
For chicken carcass breast blood-related defects (CBDs), which occur with high frequency, the visual features are approximated in terms of the similarity of the composition of these defects, making it challenging to classify them, either manually or automatically, using conventional machine vision. The aim of this paper was to introduce a method of CBD classification based on hyperspectral imaging combined with Convolutional Neural Networks (CNNs). To process hyperspectral data, the Improved Firefly Band Selection Algorithm was constructed with the 1-D CNN CBD classification model as the objective function, achieving a reduction in the dimensionality of hyperspectral data. The multidimensional data CBD classification models were developed based on YOLOv4 and Faster R-CNN, incorporating the 1-D CNN CBD classification model and the feature fusion layer. The combination of hyperspectral data and CNN can effectively accomplish the classification of CBDs, although different model architectures emphasize classification speed and accuracy differently. The multidimensional data YOLOv4 CBD classification model achieves an mAP of 0.916 with an inference time of 41.8 ms, while the multidimensional data Faster R-CNN CBD classification model, despite having a longer inference time of 58.2 ms, reaches a higher mAP of 0.990. In practical production scenarios, the appropriate classification model can be selected based on specific needs.
引用
收藏
页数:19
相关论文
共 31 条
[1]   Semisupervised dimensionality reduction for hyperspectral images based on the combination of semisupervised learning and metric learning [J].
Ahmadi, Seyyed Ali ;
Mehrshad, Nasser ;
Razavi, Seyyed Mohammad .
IMAGING SCIENCE JOURNAL, 2018, 66 (05) :320-327
[2]  
Ahn Jin-Ho, 2020, [Journal of Sensor Science and Technology, 센서학회지], V29, P427
[3]   The use of partially slatted floor designs as an alternative to littered systems in broiler chickens. I. The effects on the performance, slaughter and carcass traits [J].
Aslan, Resul ;
Sarica, Musa ;
Cavdarci, Hatice ;
Erensoy, Kadir ;
Karacay, Numan .
TROPICAL ANIMAL HEALTH AND PRODUCTION, 2024, 56 (08)
[4]   Specialized protein products in broiler chicken nutrition: A review [J].
Beski, Sleman S. M. ;
Swick, Robert A. ;
Iji, Paul A. .
ANIMAL NUTRITION, 2015, 1 (02) :47-53
[5]  
Caporaso N., 2021, INNOVATIVE FOOD ANAL, P345
[6]  
Chagas P, 2018, IEEE IJCNN
[7]  
Ghasemi M., 2022, Decis Anal J, V5, P100125, DOI [10.1016/j.dajour.2022.100125, DOI 10.1016/J.DAJOUR.2022.100125]
[8]   Fast detection and visualization of minced lamb meat adulteration using NIR hyperspectral imaging and multivariate image analysis [J].
Kamruzzaman, Mohammed ;
Sun, Da-Wen ;
ElMasry, Gamal ;
Allen, Paul .
TALANTA, 2013, 103 :130-136
[9]  
Kang R, 2016, APPL ENG AGRIC, V32, P133
[10]  
Kim T, 2010, REV COLOMB CIENC PEC, V23, P9