On-line Foreign Object Detection Method for Chicken Wing Packaging Lines Based on Improved YOLOv8

ZHANG Weiping; WANG Dongyun; JI Li; LI Guoqiang; YANG Yu

doi:10.19554/j.cnki.1001-3563.2026.05.020

PDF(3287 KB)

Packaging Engineering ›› 2026, Vol. 47 ›› Issue (5) : 181-189. DOI: 10.19554/j.cnki.1001-3563.2026.05.020

Automatic and Intelligent Technology

On-line Foreign Object Detection Method for Chicken Wing Packaging Lines Based on Improved YOLOv8

ZHANG Weiping¹, WANG Dongyun¹, JI Li¹, LI Guoqiang^2,3, YANG Yu^1,*

Author information +

History +

Abstract

The work aims to propose a lightweight on-line detection method based on an improved YOLOv8n, named M-YOLOv8 to address the high miss-detection rate of tiny non-metallic foreign objects (such as hair and plastic fibers) attached to product surfaces in fresh meat automated packaging lines, which are highly susceptible to interference from specular reflections and complex physiological textures. Firstly, an offline instance-level Copy-Paste algorithm combined with Poisson blending was employed to construct a small-target augmented dataset. This process specifically targeted "foreign object attachment" scenarios by simulating the light-shadow fusion characteristics between objects and meat surfaces to correct long-tail distribution bias. Secondly, a CBAM (Convolutional Block Attention Module) was embedded into the C2f module of the backbone to suppress fat reflections and enhance the feature-to-noise ratio in target regions. A P2 micro-scale detection layer and the SAHI (Slicing Aided Hyper Inference) strategy were integrated to compensate for spatial information loss caused by deep feature downsampling. Finally, the CIoU loss function was utilized to optimize the bounding box regression accuracy for slender linear attachments. Experimental results on a self-built chicken wing packaging dataset showed that the AP@0.5 of the proposed method reached 91.8%, a 4.3% increase over the baseline model. Specifically, the detection rate for micro-objects with a diameter of less than 2 mm was significantly improved. The system inference speed reached 98 FPS, meeting the real-time requirements of high-speed production lines. The proposed method effectively balances the detection accuracy of attached targets under complex backgrounds with edge-side inference speed. It provides a reliable technical reference for addressing the challenges of "long-tail distribution" and specular interference in meat and poultry packaging detection.

Key words

packaging detection / micro-foreign objects / YOLOv8n / SAHI / attention mechanism / Copy-Paste

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

ZHANG Weiping, WANG Dongyun, JI Li, LI Guoqiang, YANG Yu. On-line Foreign Object Detection Method for Chicken Wing Packaging Lines Based on Improved YOLOv8[J]. Packaging Engineering. 2026, 47(5): 181-189 https://doi.org/10.19554/j.cnki.1001-3563.2026.05.020

References

[1] 刘琳. 小鸡炖蘑菇预制菜的加工工艺及品质控制研究[D]. 沈阳: 沈阳农业大学, 2025.
[2] LIU L.Research on Processing Technology and Quality Control of Chicken Stewed Mushroom Prefabricated Dishes[D]. Shenyang: Shenyang Agricultural University, 2025.
[3] 王振新, 蒋亚军, 吴文龙, 等. 基于改进YOLOv8s模型的茶叶异物检测方法分析[J]. 中国机械, 2025(30): 122-128.
[4] WANG Z X, JIANG Y J, WU W L, et al.Analysis of Tea Foreign Body Detection Method Based on Improved YOLOv8s Model[J]. Machine China, 2025(30): 122-128.
[5] 王晓冰. 基于改进YOLO算法的动物源性食品检测方法[J]. 现代食品, 2024, 30(9): 91-93.
[6] WANG X B.Animal Derived Food Detection Method Based on Improved YOLO Algorithm[J]. Modern Food, 2024, 30(9): 91-93.
[7] 杨涛存. 分布外样本检测方法及其应用研究[D]. 北京: 北京交通大学, 2024.
[8] YANG T C.Out of Distribution Detection Algorithm and Application[D]. Beijing: Beijing Jiaotong University, 2024.
[9] 崔新霞, 卢硕晨, 孙敦凯. 基于视觉感知的机器人工件识别方法研究[J]. 包装工程, 2023, 44(7): 186-195.
[10] CUI X X, LU S C, SUN D K.Robot Workpiece Recognition Method Based on Visual Perception[J]. Packaging Engineering, 2023, 44(7): 186-195.
[11] 陈泳益, 蓝积炎, 杨喜, 等. 数据不均衡条件下数据增强辅助的自动调制识别[J]. 华侨大学学报(自然科学版), 2026, 47(1): 104-111.
[12] CHEN Y Y, LAN J Y, YANG X, et al.Data Augmentation-Assisted Automatic Modulation Recognition under Condition of Data Imbalance[J]. Journal of Huaqiao University (Natural Science), 2026, 47(1): 104-111.
[13] SHEN Z Y, MAO M Y, FAN P F.A Primary Comparison of Diffusion Models and Generative Adversarial Networks for Image Synthesis[C]// Proceedings of the 2024 7th International Conference on Machine Learning and Machine Intelligence (MLMI). Osaka: ACM, 2024: 225-234.
[14] CHAKRABORTY T, REDDY K S U, NAIK S M, et al. Ten Years of Generative Adversarial Nets (GANs): A Survey of the State-of-the-Art[J]. Machine Learning: Science and Technology, 2024, 5(1): 011001.
[15] ZHANG C, CHEN J Y, WEN Y B, et al.High-Efficiency Γ-Aminobutyric Acid Nano-Formulation for Turnip Mosaic Virus: Excellent Adhesion Performance and Amplified Plant Defensive Responses[J]. Horticultural Plant Journal, 2025, 11(6): 2285-2288.
[16] 李英, 陈健, 苏志海, 等. 基于Mixup训练及多模型决策融合的腰椎间盘突出诊断[J]. 信息工程大学学报, 2024, 25(3): 265-271.
[17] LI Y, CHEN J, SU Z H, et al.Diagnosis of Lumbar Disc Herniation Based on Mixup Training and Decision Fusion of Multiple Models[J]. Journal of Information Engineering University, 2024, 25(3): 265-271.
[18] 严梦迪, 程旭, 付章杰. 基于多尺度特征的图像复制粘贴篡改检测网络[J]. 计算机与数字工程, 2025, 53(10): 2881-2887.
[19] YAN M D, CHENG X, FU Z J.Multi-Scale Feature Network for Image Copy-Move Forgery Detection[J]. Computer and Digital Engineering, 2025, 53(10): 2881-2887.
[20] 江松, 魏玉, 饶彬舰, 等. 融合Copy-Paste的大尺度露天矿岩质边坡裂隙检测方法研究[J]. 采矿与岩层控制工程学报, 2025, 7(5): 144-159.
[21] JIANG S, WEI Y, RAO B J, et al.A Large-Scale Open-Pit Rock Slope Fissure Detection Method by Integrating Copy-Paste Algorithm[J]. Journal of Mining and Strata Control Engineering, 2025, 7(5): 144-159.
[22] VARGHESE R, M S. YOLOv8: A Novel Object Detection Algorithm with Enhanced Performance and Robustness[C]// 2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS). Chennai: IEEE, 2024.
[23] 白雄飞, 龚水成, 李雪松, 等. 基于泊松融合数据增强的焊缝金相组织缺陷分类研究[J]. 上海交通大学学报, 2023, 57(10): 1316-1328.
[24] BAI X F, GONG S C, LI X S, et al.Defect Classification of Weld Metallographic Structure Based on Data Augmentation of Poisson Fusion[J]. Journal of Shanghai Jiao Tong University, 2023, 57(10): 1316-1328.
[25] LIN S M, JIANG Y, CHEN X S, et al.Automatic Detection of Plant Rows for a Transplanter in Paddy Field Using Faster R-CNN[J]. IEEE Access, 2020, 8: 147231-147240.
[26] 李晓光, 付陈平, 李晓莉, 等. 面向多尺度目标检测的改进Faster R-CNN算法[J]. 计算机辅助设计与图形学学报, 2019, 31(7): 1095-1101.
[27] LI X G, FU C P, LI X L, et al.Improved Faster R-CNN for Multi-Scale Object Detection[J]. Journal of Computer-Aided Design & Computer Graphics, 2019, 31(7): 1095-1101.
[28] LIU W, ANGUELOV D, ERHAN D, et al.SSD: Single Shot MultiBox Detector[C]// Computer Vision—ECCV 2016. Cham: Springer, 2016.
[29] 王军, 万书东, 程勇. 基于改进YOLOv5s的白酒瓶盖瑕疵检测[J]. 包装工程, 2024, 45(7): 180-188.
[30] WANG J, WAN S D, CHENG Y.Liquor Bottle Cap Defect Detection Based on Improved YOLOv5s[J]. Packaging Engineering, 2024, 45(7): 180-188.
[31] 杨明旭, 张俊宁, 张志强, 等. 基于改进YOLOv8的药片泡罩包装缺陷检测算法[J]. 包装工程, 2025, 46(1): 145-154.
[32] YANG M X, ZHANG J N, ZHANG Z Q, et al.Defect Detection Algorithm of Pharmaceutical Blister Package Based on Improved YOLOv8[J]. Packaging Engineering, 2025, 46(1): 145-154.
[33] 徐淼, 涂福泉, 吴淇, 等. 基于YOLOv8的PCB表面缺陷检测轻量化研究[J]. 包装工程, 2024, 45(17): 172-179.
[34] XU M, TU F Q, WU Q, et al.Lightweight Research on PCB Surface Defect Detection Based on YOLOv8[J]. Packaging Engineering, 2024, 45(17): 172-179.