基于改进YOLOv10的喷码微小字符精确定位算法

doi:10.3778/j.issn.1002-8331.2410-0482

摘要/Abstract

摘要： 为解决自动化生产领域物品包装表面喷码字符因微小尺寸及特征模糊而导致的检测定位精度差问题，将YOLOv10n作为基线网络提出一种针对特征模糊小目标检测的精确定位算法（YOLO-DLW）。使用细节信息提取卷积（detail information extraction convolution，DIEConv）模块替换基线网络中所有的跨步卷积模块，避免其导致的细节特征丢失问题，提高网络对小目标特征的提取能力。引入低级特征融合检测层，减少基线网络在下采样过程中造成的小目标特征损失。在颈部网络采用加权混合融合金字塔网络（weighted hybrid fusion pyramid network，WHFPN）结构，并结合内容引导注意力（content-guided attention，CGA）机制，有效提升特征层间的信息融合效率和网络对关键信息的提取能力。与基线模型相比，YOLO-DLW算法应用在编织袋小目标字符定位数据集上，准确率、召回率、mAP@0.5和mAP@0.5：0.95分别提高了7.3、8.2、3.9和3.3个百分点，有效解决基线模型对字符区域的误检和漏检问题。

关键词: 小目标检测, 精确定位, YOLOv10n, 细节信息提取卷积, 加权混合融合金字塔, 内容引导注意力

Abstract: To address the issue of poor detection and localization accuracy of inkjet characters on product packaging surfaces in automated production, due to their small size and ambiguous features, a precise localization algorithm (YOLO-DLW) for detecting small targets with ambiguous features is proposed, employing YOLOv10n as the baseline network. All strided convolution modules in the baseline network are replaced with detail information extraction convolution (DIEConv) modules to prevent the loss of detail features and enhance the network’s capability to extract features from small targets. A low-level feature fusion detection layer is introduced to reduce the loss of small target features caused by the downsampling process in the baseline network. In the neck network, a weighted hybrid fusion pyramid network (WHFPN) structure is employed, combined with the content-guided attention (CGA) mechanism, to effectively enhance the efficiency of information fusion between feature layers and the network’s ability to extract critical information. Compared to the baseline model, the YOLO-DLW algorithm, when applied to the woven bag small target character localization dataset, shows improvements of 7.3, 8.2, 3.9, and 3.3 percentage points in accuracy, recall rate, mAP@0.5, and mAP@0.5：0.95 respectively, effectively resolving the issues of false positives and missed detections in the character area of the baseline model.

Key words: small targets detection, precise localization, YOLOv10n, detail information extraction convolution, weighted hybrid fusion pyramid, content-guided attention

操振, 余朝刚, 靳胜洁, 王帅鹏, 朱文良. 基于改进YOLOv10的喷码微小字符精确定位算法[J]. 计算机工程与应用, 2025, 61(7): 153-164.

CAO Zhen, YU Chaogang, JIN Shengjie, WANG Shuaipeng, ZHU Wenliang. Precise Localization Algorithm for Inkjet Small Characters Based on Improved YOLOv10[J]. Computer Engineering and Applications, 2025, 61(7): 153-164.

参考文献

[1] MELEK C G, BATTINI S?NMEZ E, VARLI S. Datasets and methods of product recognition on grocery shelf images using computer vision and machine learning approaches: an exhaustive literature review[J]. Engineering Applications of Artificial Intelligence, 2024, 133: 108452.
[2] GINER J, KATIC D, KOVACS K, et al. A computer vision based approach to reduce system downtimes in an automated high-rack logistics warehouse[J]. Procedia CIRP, 2023, 118: 1078-1083.
[3] YAN J H, WANG Z P. YOLO V3 + VGG16-based automatic operations monitoring and analysis in a manufacturing workshop under Industry 4.0[J]. Journal of Manufacturing Systems, 2022, 63: 134-142.
[4] MIRBOD M, GHATARI A R, SAATI S, et al. Industrial parts change recognition model using machine vision, image processing in the framework of industrial information integration[J]. Journal of Industrial Information Integration, 2022, 26: 100277.
[5] 颜豪男, 吕伏, 冯永安. 特征级自适应增强的无人机目标检测算法[J]. 计算机科学与探索, 2024, 18(6): 1566-1578.
YAN H N, LYU F, FENG Y A. Feature-level adaptive enhancement for UAV target detection algorithm[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(6): 1566-1578.
[6] WANG T Y, HARDIN R G I, WARD J K, et al. A smart cotton module tracking and monitoring system for handling logistics and cover damage[J]. Computers and Electronics in Agriculture, 2022, 193: 106620.
[7] DI CAPUA M, CIARAMELLA A, DE PRISCO A. Machine learning and computer vision for the automation of processes in advanced logistics: the integrated logistic platform(ILP) 4.0[J]. Procedia Computer Science, 2023, 217: 326-338.
[8] 徐彦威, 李军, 董元方, 等. YOLO系列目标检测算法综述[J]. 计算机科学与探索, 2024, 18(9): 2221-2238.
XU Y W, LI J, DONG Y F, et al. Survey of development of YOLO object detection algorithms[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(9): 2221-2238.
[9] 葛永杰, 王丽丹, 陈定喜, 等. 基于深度学习的工业视觉箱体字符识别与判断[J]. 计算机工程, 2022, 48(1): 296-304.
GE Y J, WANG L D, CHEN D X, et al. Character recognition and judgment of industrial vision box based on deep learning[J]. Computer Engineering, 2022, 48(1): 296-304.
[10] 高磊, 徐洪, 汪昌平. 基于改进Yolo v3的H型钢喷号识别系统研究[J]. 冶金动力, 2024, 43(3): 63-66.
GAO L, XU H, WANG C P. Research on H-beam spray number identification system based on improved Yolo v3[J]. Metallurgical Power, 2024, 43(3): 63-66.
[11] 马宏伟. 平板工件字符编码视觉识别技术研究[D]. 哈尔滨: 哈尔滨商业大学, 2024.
MA H W. Research on visual recognition technology for character encoding of flat workpieces[D]. Harbin: Harbin University of Commerce, 2024.
[12] ZHANG X S, WANG Y. Industrial character recognition based on improved CRNN in complex environments[J]. Computers in Industry, 2022, 142: 103732.
[13] 周玮, 门耀华, 辛立刚. 基于机器视觉的柔性包装袋喷码缺陷检测研究[J]. 包装工程, 2022, 43(9): 249-256.
ZHOU W, MEN Y H, XIN L G. Inspection of coding defects in flexible packaging bags based on machine vision[J]. Packaging Engineering, 2022, 43(9): 249-256.
[14] 劳昶皓. 基于深度学习的喷码检测识别系统研究[D]. 桂林: 广西师范大学, 2023.
LAO C H. Research on coding detection and recognition system based on deep learning[D]. Guilin: Guangxi Normal University, 2023.
[15] 李帆, 胡维平, 刘北北, 等. 基于NVIDIA TX2的喷码字符检测算法[J]. 计算机工程与应用, 2022, 58(13): 210-216.
LI F, HU W P, LIU B B, et al. NVIDIA TX2-based inkjet character detection algorithm[J]. Computer Engineering and Applications, 2022, 58(13): 210-216.
[16] 宋存利, 柴伟琴, 张雪松. 基于改进YOLO v5算法的道路小目标检测[J]. 系统工程与电子技术, 2024, 46(10): 3271-3278.
SONG C L, CHAI W Q, ZHANG X S. Road small target detection based on improved YOLO v5 algorithm[J]. Systems Engineering and Electronics, 2024, 46(10): 3271-3278.
[17] LI X M, WEI Y K, LI J H, et al. Improved YOLOv7 algorithm for small object detection in unmanned aerial vehicle image scenarios[J]. Applied Sciences, 2024, 14(4): 1664.
[18] 王呈, 王炀, 荣英佼. 面向配电柜字符识别的YOLOv7-MSBP目标定位算法[J]. 计算机应用, 2024, 44(10): 3191-3199.
WANG C, WANG Y, RONG Y J. YOLOv7-MSBP target location algorithm for character recognition of power distribution cabinet[J]. Journal of Computer Applications, 2024, 44(10): 3191-3199.
[19] WANG F, WANG H Y, QIN Z Y, et al. UAV target detection algorithm based on improved YOLOv8[J]. IEEE Access, 2023, 11: 116534-116544.
[20] WANG H, HAN D Z, CUI M M, et al. NAS-YOLOX: a SAR ship detection using neural architecture search and multi-scale attention[J]. Connection Science, 2023, 35(1): 1-32.
[21] XU W Y, CUI C, JI Y C, et al. YOLOv8-MPEB small target detection algorithm based on UAV images[J]. Heliyon, 2024, 10(8): e29501.
[22] 王传云, 苏阳, 王琳霖, 等. 面向反制无人机集群的多目标连续鲁棒跟踪算法[J]. 航空学报, 2024, 45(7): 261-274.
WANG C Y, SU Y, WANG L L, et al. Multi-object continuous robust tracking algorithm for anti-UAV swarm[J]. Acta Aeronautica et Astronautica Sinica, 2024, 45(7): 261-274.
[23] SUNKARA R, LUO T. No more strided convolutions or pooling: a new CNN building block for low-resolution images and small objects[C]//Proceedings of the Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Cham: Springer, 2023: 443-459.
[24] HAN K, WANG Y H, TIAN Q, et al. GhostNet: more features from cheap operations[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 1577-1586.
[25] VARGHESE R, SAMBATH M. YOLOv8: a novel object detection algorithm with enhanced performance and robustness[C]//Proceedings of the 2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems. Piscataway: IEEE, 2024: 1-6.
[26] TAN M X, PANG R M, LE Q V. EfficientDet: scalable and efficient object detection[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 10778-10787.
[27] CHEN Z X, HE Z W, LU Z M. DEA-Net: single image dehazing based on detail-enhanced convolution and content-guided attention[J]. IEEE Transactions on Image Processing, 2024, 33: 1002-1015.
[28] YANG G Y, LEI J, ZHU Z K, et al. AFPN: asymptotic feature pyramid network for object detection[C]//Proceedings of the 2023 IEEE International Conference on Systems, Man, and Cybernetics. Piscataway: IEEE, 2023: 2184-2189.
[29] XU X Z, JIANG Y Q, CHEN W H, et al. DAMO-YOLO: a report on real-time object detection design[J]. arXiv:2211. 15444, 2022.
[30] HSU P H, LEE P J, BUI T A, et al. YOLO-SPD: tiny objects localization on remote sensing based on You only look once and space-to-depth convolution[C]//Proceedings of the 2024 IEEE International Conference on Consumer Electronics. Piscataway: IEEE, 2024: 1-3.
[31] WANG C Y, YEH I H, LIAO H M. YOLOv9: learning what you want to learn using programmable gradient information[J]. arXiv:2402.13616, 2024.
[32] WANG A, CHEN H, LIU L H, et al. YOLOv10: real-time end-to-end object detection[J]. arXiv:2405.14458, 2024.