Application of Hypergraph-Driven "Trinity" Semantic Framework in Retail Commodity Detection

CAO ZHIGANG; LIU LIBO; LIU QIAN; DENG ZHEN

doi:null

您当前的位置：

首页 >

文章列表页 >

Application of Hypergraph-Driven "Trinity" Semantic Framework in Retail Commodity Detection

更新时间：2026-03-03

- Application of Hypergraph-Driven "Trinity" Semantic Framework in Retail Commodity Detection
- Chinese Journal on Internet of Things (2026)
- 作者机构：
  
  1.宁夏大学信息工程学院，宁夏银川 750021
  2.宁夏“东数西算”人工智能与信息安全重点实验室，宁夏银川 750021
- 作者简介：
- 基金信息：
  
  The 2024 Key Research and Development Plan (Talent Introduction) Project of Ningxia(2024BEH04026)
- DOI：
  CLC： TP18
- Received：19 September 2025，
  
  Revised：2025-11-10，
  
  Accepted：26 November 2025，
- 稿件说明：
移动端阅览
CAO ZHIGANG, LIU LIBO, LIU QIAN, et al. Application of Hypergraph-Driven "Trinity" Semantic Framework in Retail Commodity Detection[J/OL]. Chinese Journal on Internet of Things, 2026.
DOI：

CAO ZHIGANG, LIU LIBO, LIU QIAN, et al. Application of Hypergraph-Driven "Trinity" Semantic Framework in Retail Commodity Detection[J/OL]. Chinese Journal on Internet of Things, 2026. DOI：

摘要

零售自助结账系统的运行效率直接影响消费者购物体验。现有系统在提取小商品的细粒度关键特征方面存在困难，同时在严重遮挡场景中空间推理能力不足，限制了检测性能的提升。为解决上述问题，本文在YOLOv12的基础上设计超图驱动的三位一体语义框架。该框架首先通过构建的跨层语义收集模块，结合跨层特征注意力融合子模块（CFAF，Cross-Layer Feature Attention Fusion）与风车卷积（PSConv，Pinwheel-Shaped Convolution），实现跨层级语义互补与深度网络中细粒度特征的高效保留，从而增强对小商品细粒度特征的捕捉能力。其次，引入语义精炼模块，其核心混合动态感知的超图计算（HDP-HGC，Hybrid Dynamic Perception Hypergraph Computation）能有效提升了系统在严重遮挡场景下的识别性能。最后，设计语义映射模块，搭建了非相邻层语义交互通道，三个模块协同构成完整的三位一体语义框架。我们在公开数据集RPC、FinVolution以及真实场景下自建数据集Mine进行了系统评估。实验结果表明，本文算法在平均检测精度上达到了90.94%、92.04%和69.61%，领先目前大多主流模型，特别是在小商品识别和严重遮挡结算场景中，表现出显著优势，在保持较高识别准确率的同时有效控制了模型参数。此外，本文方法将漏检和误检概率降低了约30%和50%，进一步验证了其在复杂零售环境下的稳定性与实用价值。

Abstract

The efficiency of retail checkout systems directly affects the shopping experience. Current systems have difficulty capturing key details of Small-sized merchandise and lack spatial reasoning in crowded scenes

which limits detection performance. To address the aforementioned issues

this paper develops a hypergraph-driven trinity semantic framework based on YOLOv12. First

the framework uses a cross-layer semantic gathering module. It combines the Cross-Layer Feature Attention Fusion (CFAF) submodule with Pinwheel-Shaped Convolution (PSConv). This process enables complementary semantics across different network layers. It also effectively preserves fine-grained features in the deep network. As a result

the model's ability to capture detailed features of small commodities is significantly improved. Next

a semantic refinement module is introduced. Its core component

Hybrid Dynamic Perception Hypergraph Computation (HDP-HGC)

effectively enhances the system's recognition performance in heavily occluded scenarios. Finally

a semantic mapping module is designed to establish semantic interaction pathways between non-adjacent layers. These three modules work collaboratively to form a complete trinity semantic framework. We conducted systematic evaluations on the public datasets RPC

FinVolution

and our self-built real-world dataset Mine. Experimental results show that our method achieved average detection accuracies of 90.94%

92.04%

and 69.61%

respectively—surpassing most mainstream models. In particular

it demonstrates significant advantages in small-object recognition and heavily occluded checkout scenarios

maintaining high accuracy while effectively controlling model parameters. Furthermore

our approach reduces missed detections and false detections by approximately 30% and 50%

respectively

further verifying its stability and practical value in complex retail environments.

关键词

Keywords

references

Janiesch C. , Zschech P. , Heinrich K. Machine Learning and Deep Learning [J ] . Electronic Markets , 2021 , 31 ( 3 ): 685 – 695 .

Zou Z. X. , Chen K. Y. , Shi Z. W. , Guo Y. H. , Ye J. P. Object Detection in 20 Years: A Survey[J ] . Proceedings of the IEEE , 2023 , 111 ( 3 ): 257 – 276 .

Wu P. , Zhang Z. , Peng X. , Wang R. Deep Learning Solutions for Smart City Challenges in Urban Development [J ] . Scientific Reports , 2024 , 14 ( 1 ): 5176 .

张静 , 农昌瑞 , 杨智勇 . 基于卷积神经网络的目标检测算法综述 [J ] . 兵器装备工程学报 , 2022 , 43 ( 06 ): 37 - 47 .

夏檑 , 袁海兵 , 吴俊 . 基于改进Faster RCNN的汽车管件密封圈装配检测研究 [J ] . 计算机应用与软件 , 2025 , 42 ( 06 ): 93 - 99 .

王伟辉 , 信泽阳 , 车清论 , 等 . 基于改进Faster RCNN模型的冬枣缺陷检测方法 [J ] . 农业工程学报 , 2024 , 40 ( 22 ): 283 - 289 .

吕晓华 , 魏铭辰 , 刘立波 . 基于位置可学习视觉中心机制的零售商品检测方法 [J ] . 物联网学报 , 2023 , 7 ( 04 ): 142 - 152 .

马超伟 , 张浩 , 马新明 , 等 . 基于改进YOLOv8的轻量化小麦病害检测方法 [J ] . 农业工程学报 , 2024 , 40 ( 05 ): 187 - 195 .

Sun S. , Mo B. , Xu J. , Li D. , Zhao J. , Han S. Multi-YOLOv8: An Infrared Moving Small Object Detection Model Based on YOLOv 8 for Air Vehicle[J ] . Neurocomputing , 2024 , 588 : 127685 .

周昶雯 , 宋强 , 张月 . 基于改进YOLOv8的凝视雷达小目标检测算法 [J ] . 信号处理 , 2025 , 41 ( 05 ): 853 - 866 .

雷帮军 , 余翱 , 吴正平 , 等 . 改进YOLOv8n的无人机航拍小目标检测算法 [J ] . 现代电子技术 , 2025 , 48 ( 03 ): 26 - 34 . DOI: 10.16652/j.issn.1004-373x.2025.03.005 http://dx.doi.org/10.16652/j.issn.1004-373x.2025.03.005 .

Wu S. , Lu X. , Guo C. , Guo H. MV-YOLO: An Efficient Small Object Detection Framework Based on Mamba [J ] . IEEE Transactions on Geoscience and Remote Sensing , 2025 , 63 : 5632814 .

冯庆胜 , 付明雨 , 姚泽圆 , 等 . 基于改进YOLOv8的轨道小尺度异物入侵算法研究 [J ] . 现代电子技术 , 2025 , 48 ( 11 ): 174 - 179 . DOI: 10.16652/j.issn.1004-373x.2025.11.027 http://dx.doi.org/10.16652/j.issn.1004-373x.2025.11.027 .

Ye J. , Zhang Y. , Li P. , Guo Z. , Zeng S. , Wei T. Real-Time Dense Small Object Detection Model for Floating-Litter Detection and Removal on Water Surfaces [J ] . Marine Pollution Bulletin , 2025 , 218 : 118189 .

罗偲 , 李凯扬 , 吴吉花 , 等 . 基于对抗注意力机制的水下遮挡目标检测算法 [J ] . 计算机工程 , 2024 , 50 ( 10 ): 313 - 321 .

Tian Y. , Ye Q. , Doermann D. YOLOv12: Attention-Centric Real-Time Object Detectors . arXiv preprint : arXiv: 2502.12524 , 2025 .

Dao T. , Fu D. , Ermon S. , Rudra A. , Re, C. FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness [C ] // Advances in Neural Information Processing Systems . Vancouver, Canada : Curran Associates, Inc , 2022 , 35 : 1 6344– 16359 .

Lin T. Y. , Dollár P. , Girshick R. , He K. , Hariharan B. , Belongie, S. Feature Pyramid Networks for Object Detection [C ] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition . Honolulu, HI, USA : IEEE Computer Society , 2017 : 2117 - 2125 .

Xu S. , Zheng S. , Xu W. , Xu R. , Wang C. , Zhang J. , Guo, L. HEF-Net: Hierarchical Context Fusion Network for Infrared Small Object Detection [C ] // 2024 IEEE International Conference on Multimedia and Expo (ICME) . Singapore : IEEE , 2024 : 1 – 6 .

Varghese R. , Sambath , M. YOLOv 8 : 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, India : IEEE , 2024: 1 - 6 .

Yang J. , Liu S. , Wu J. , Su X. , Hai N. , Huang, X. Pinwheel-Shaped Convolution and Scale-Based Dynamic Loss for Infrared Small Target Detection [C ] // Proceedings of the AAAI Conference on Artificial Intelligence . Philadelphia, Pennsylvania, USA : AAAI Press , 2025 , 39 : 9202 – 9210 .

Bai S. , Zhang F. , Torr P. H. Hypergraph Convolution and Hypergraph Attention [J ] . Pattern Recognition , 2021 , 110 : 107637 .

Cai D. , Song M. , Sun C. , Zhang B. , Hong S. , Li, H. Hypergraph Structure Learning for Hypergraph Neural Networks [C ] // Proceedings of the 31st International Joint Conference on Artificial Intelligence (IJCAI-22) . Vienna, Austria : IJCAI Press , 2022 : 1923 – 1929 .

Chien E. , Pan C. , Peng J. , Milenkovic, O. You Are AllSet: A Multiset Function Framework for Hypergraph Neural Networks [C ] // Proceedings of the Tenth International Conference on Learning Representations (ICLR 2022) . Virtual : ICLR Organization , 2022 .

Feng Y. , Huang J. , Du S. , Ying S. , Yong J. H. , Li Y. , Gao Y. Hyper-YOLO: When Visual Object Detection Meets Hypergraph Computation [J ] . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025 , 47 ( 4 ): 2388 - 2401 .

Wei X. S. , Cui Q. , Yang L. , Wang P. , Liu L. Q. , Yang J. RPC: A Large-Scale and Fine-Grained Retail Product Checkout Dataset [J ] . Science China Information Sciences , 2022 , 65 ( 9 ): 1 – 2 .

Carion N. , Massa F. , Synnaeve G. , Usunier N. , Kirillov A. , Zagoruyko, S. End-to-End Object Detection with Transformers [C ] // Proceedings of the European Conference on Computer Vision (ECCV) . Virtual : Springer , 2020 : 213 – 229 .

Huang S. , Lu Z. , Cun X. , Yu Y. , Zhou X. , Shen, X. DEIM: DETR with Improved Matching for Fast Convergence [C ] // Proceedings of the IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR) . Nashville, Tennessee, USA : IEEE , 2025 : 15162 – 15171 .

Ding X. , Zhang X. , Han J. , Ding G. Scaling Up Your Kernels to 31 × 31 : Revisiting Large Kernel Design in CNNs [C ] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . New Orleans, LA, USA : IEEE Computer Society , 2022: 11963 – 11975 .

Narkhede J. ( 2024 ). Comparative evaluation of post-hoc explainability methods in AI: LIME, SHAP, and Grad-CAM [C ] // 2024 4th International Conference on Innovative Computing, Intelligent Communication and Smart Electrical Systems (ICSES) , Guilin, China : IEEE , 2024 : 826 - 830 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Spectrum allocation and device association in ultra-dense industrial IoT via hypergraph matching

Service relationship network modeling and service composition optimization algorithm based on hypergraph

Related Author

WANG Qin

CHANG Haotian

ZHU Hongbo

JIANG Lingyun

YANG Xuewei

Related Institution

Institute of Internet of Things, Nanjing University of Posts and Telecommunications

Jiangsu Key Laboratory of Wireless Communications and Internet of Things

Engineering Research Center of Health Service System based on Ubiquitous Wireless Networks

College of Telecommunications and Information Engineering, Nanjing University of Posts and Telecommunications

⁰