LSGI-YOLOv8:基于轻量化YOLOv8的瓷砖表面缺陷检测算法

doi:10.3969/j.issn.1006-2475.2025.11.010

摘要/Abstract

摘要： 摘要：针对瓷砖表面缺陷小、花纹复杂，造成瓷砖缺陷检测精度低、速度慢等问题，本文提出一种基于改进YOLOv8的瓷砖表面缺陷检测算法。该算法设计一种融合LSKA （Large Separable Kernel Attention）注意力机制的空间池化层模块LSPPF，提高模型的多尺度特征提取能力；设计一种轻量级特征融合网络SimBiFPN，消减加权双向特征金字塔网络中的跳跃连接次数，加强模型对小目标缺陷特征的表示，并减少模型的参数量；提出一种基于新的轻量级卷积GSConv （Grouped Shuffle Convolution）和深度可分离卷积（Depthwise Separable Convolution， DSConv）的D-Slim-Neck模块，用于替换传统Slim-Neck模块，显著降低模型计算量和复杂度；针对瓷砖表面缺陷形状不一的问题，提出一种带有辅助框的损失函数Inner_CIoU，通过调整尺度因子来控制辅助框的大小，进一步提高算法的检测精度。实验结果表明，改进后算法的平均精度均值达到90.3%，比YOLOv8提高2.1百分点，参数量和计算量分别减少6.7%和13.6%，并达到了与其他先进算法相媲美的瓷砖表面缺陷检测性能。

关键词: 关键词：瓷砖缺陷检测, YOLOv8, 大核注意力机制, 多尺度特征融合, GSConv

Abstract: Abstract: Aiming at the problem of low detection accuracy and slow detection speed of ceramic tile defect detection caused by small defects and complex patterns on the surface of ceramic tiles， a ceramic tile surface defect detection algorithm based on improved YOLOv8 is proposed. The algorithm designs a spatial pooling layer module LSPPF incorporating the LSKA （Large Separable Kernel Attention） mechanism to improve the multi-scale feature extraction capability of the model； a lightweight feature fusion network SimBiFPN is designed to reduce the number of jump connections in the weighted bidirectional feature pyramid network， enhance the model’s representation of small target defect features， and reduce the number of model parameters; a D-Slim-Neck module based on lightweight convolution GSConv （Grouped Shuffle Convolution） and depth-separable convolution （Depthwise Separable Convolution， DSConv） is proposed for replacing the traditional Slim-Neck module to significantly reduce the model computation and complexity; to address the problem of different shapes of tile surface defects， a loss function Inner_CIoU with an auxiliary box is proposed to control the size of the auxiliary frame by adjusting the scale factor to further improve the detection accuracy of the algorithm. The experimental results show that the mean average precision of the improved algorithm reaches 90.3%， which is 2.1 percentage points higher than YOLOv8， and the number of parameters and calculations are reduced by 6.7% and 13.6% respectively， achieving tile surface defect detection performance comparable to other advanced algorithms.

Key words: Key words: ceramic tile defect detection, YOLOv8, large kernel attention mechanism, multi-scale feature fusion, GSConv

中图分类号:

杨安博1, 钟国韵1, 刘梅锋2, 习超2, 张维1, 丁鹏1. LSGI-YOLOv8:基于轻量化YOLOv8的瓷砖表面缺陷检测算法[J]. 计算机与现代化, 2025, 0(11): 80-88.

YANG Anbo1, ZHONG Guoyun1, LIU Meifeng2, XI Chao2, ZHANG Wei1, DING Peng1. LSGI-YOLOv8: Ceramic Tile Surface Defect Detection Algorithm Based on Lightweight YOLOv8[J]. Computer and Modernization, 2025, 0(11): 80-88.

参考文献

［1］肖克来提. 表面缺陷检测应用研究综述［J］. 电子技术， 2020，49（8）:189-191.
［2］ LIAN J， JIA W K， ZAREAPOOR M， et al. Deep-learning-based small surface defect detection via an exaggerated local variation-based generative adversarial network［J］. IEEE Transactions on Industrial Informatics， 2019，16（2）:1343-1351.
［3］李军华，权小霞，汪宇玲. 多特征融合的瓷砖表面缺陷检测算法研究［J］. 计算机工程与应用， 2020， 56（15）:191-198.
［4］ RAGAB K， ALSHARAY N. Developing parallel cracks and spots ceramic defect detection and classification algorithm using CUDA［C］// 2017 IEEE 13th International Symposium on Autonomous Decentralized System （ISADS）. IEEE， 2017:255-261.
［5］ MATIC T， ALEKSI I， HOCENSKI Ž， et al. Real-time biscuit tile image segmentation method based on edge detection［J］. ISA Transactions， 2018，76:246-254.
［6］ ZHANG H L， PENG L， LEI G. Saliency detection for surface defects of ceramic tile［J］. Ceramics International，2022，48（21）:32113-32124.
［7］ LIU W， ANGUELOV D， ERHAN D， et al. SSD: Single shot multibox detector［C］// Computer Vision-ECCV 2016. Springer，2016: 21-37.
［8］周曼，吴天钊，代宝鑫，等. 深度学习在陶瓷表面缺陷检测方面的研究进展［J］. 陶瓷学报， 2023，44（5）:874-884.
［9］ REDMON J，DIVVALA S，GIRSHICK R，et al. You only look once: Unified，real-time object detection［C］// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. IEEE， 2016:779-788.
［10］ GE Z，LIU S T，WANG F，et al. YOLOx: Exceeding YOLO series in 2021［J］. arXiv preprint arXiv:2107.08430， 2021.
［11］ WANG C Y，BOCHKOVSKIY A，LIAO H Y M. YOLOv7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors［C］// Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE， 2023:7464-7475.
［12］ GIRSHICK R. Fast R-CNN［C］// Proceedings of the IEEE International Conference on Computer Vision. IEEE，2015: 1440-1448.
［13］ HE K， GKIOXARI G， DOLLAR P， et al. Mask R-CNN［C］// Proceedings of the 2017 IEEE International Conference on Computer Vision. IEEE， 2017:2961-2969.
［14］赵楚，段先华，苏俊楷. 改进Faster RCNN的瓷砖表面瑕疵检测研究［J］. 计算机工程与应用， 2023， 59（14）:201-208.
［15］ LUO J X， YANG Z Y， LI S P， et al. FPCB surface defect detection: A decoupled two-stage object detection framework［J］. IEEE Transactions on Instrumentation and Measurement， 2021，70. DOI: 10.1109/TIM.2021.3092510.
［16］ ÜZEN H， TURKOGLU M， ASLAN M， et al. Depth-wise squeeze and excitation block-based efficient-unet model for surface defect detection［J］. The Visual Computer， 2023，39（5）: 1745-1764.
［17］ CAO T L， SONG K C， XU L K， et al. Balanced multi-scale target score network for ceramic tile surface defect detection［J］. Measurement， 2024，224:113914.
［18］余松森，林智凡，薛国鹏，等. 基于改进YOLOv8的轻量级大幅面瓷砖缺陷检测算法［J］. 计算机应用， 2025，45（2）:647-654.
［19］ WANG G， CHEN Y F， AN P， et al. UAV-YOLOv8: A small-object-detection model based on improved YOLOv8 for UAV aerial photography scenarios［J］. Sensors， 2023， 23（16）: 7190.
［20］ WAN G， FANG H B， WANG D Z， et al. Ceramic tile surface defect detection based on deep learning［J］. Ceramics International， 2022，48（8）:11085-11093.
［21］ LAU K W， PO L M， REHMAN Y A U. Large separable kernel attention: Rethinking the large kernel attention design in CNN［J］. Expert Systems with Applications， 2024， 236: 121352.
［22］ GUO M H， LU C Z， LIU Z N， et al. Visual attention network［J］. Computational Visual Media， 2023，9（4）:733-752.
［23］张泽，张建权，周国鹏. 基于改进YOLOv8s的摄像头模组缺陷检测［J］. 计算机与现代化， 2024（9）:107-113.
［24］ HE L， WEI H J， WANG Q X. A new target detection method of ferrography wear particle images based on ECAM-YOLOv5-BiFPN network［J］. Sensors， 2023，23（14）: 6477.
［25］ LI H L， LI J， WEI H B， et al. Slim-neck by GSConv: A better design paradigm of detector architectures for autonomous vehicles［J］. arXiv preprint arXiv:2206.02424， 2022.
［26］ ZHANG H， XU C， ZHANG S J. Inner-IoU: More effective intersection over union loss with auxiliary bounding box［J］. arXiv preprint arXiv:2311.02877， 2023.
［27］余松森，薛国鹏，何皇，等. 改进YOLOv8的轻量级瓷砖表面缺陷检测［J］. 计算机工程与应用， 2024，60（18）:88-102.
［28］ YU X L， YU Q C， ZHANG Y， et al. Research on tile surface defect detection by improved YOLOv5［J］. Journal of Intelligent & Fuzzy Systems， 2023，45（6）:11319-11331.

[1]	孙二杰1, 张奇丰2, 王德清3. 基于轻量级YOLOv8的变电站设备缺陷检测[J]. 计算机与现代化, 2025, 0(11): 65-70.
[2]	蒋冬梅1, 杨诺2, 陈仁明2, 董昌武3, 彭成东2, 4. 基于变核卷积的HRNetV2模型对舌下络脉图像分割算法[J]. 计算机与现代化, 2025, 0(10): 14-19.
[3]	段警韦1, 陈亮1, 李雪1, 刘蒙蒙1, 刘晋宇2. 改进YOLOv8的密集行人检测算法[J]. 计算机与现代化, 2025, 0(10): 24-31.
[4]	姚杰1, 殷洪海1, 汪大海1, 李润梓2, 张茜雯2, 郭毓2. 配网旁路作业机器人目标位姿估计方法[J]. 计算机与现代化, 2025, 0(09): 20-26.
[5]	罗德艳, 徐杨, 左锋云, 张永丹. 改进YOLOv8的轻量化垃圾分类检测[J]. 计算机与现代化, 2025, 0(09): 35-42.
[6]	贾涛1, 吴月超2, 吕洋2, 付文龙3. 基于改进YOLOv8-EDD的光伏板缺陷检测[J]. 计算机与现代化, 2025, 0(09): 43-49.
[7]	苏衍森, 牟莉. 改进YOLOv8的课堂行为检测算法[J]. 计算机与现代化, 2025, 0(08): 76-81.
[8]	吴栋, 范永胜, 桑彬彬. 基于改进RT-DETR的轻量级火焰检测算法[J]. 计算机与现代化, 2025, 0(07): 106-111.
[9]	李子贺, 王一丁. 不同视角三维人体关键点动作相似度计算[J]. 计算机与现代化, 2025, 0(07): 63-68.
[10]	郝荣荣1, 马巧梅1, 谭亚军1, 石桓印2. 基于可变形卷积与注意力机制的复杂火灾检测[J]. 计算机与现代化, 2025, 0(06): 79-85.
[11]	唐锐1, 武建超1, 陈剑波1, 柴江1, 王迁1, 何雨辰2. 基于GiraffeDet的改进YOLOv8s输电线路覆冰检测[J]. 计算机与现代化, 2025, 0(03): 6-11.
[12]	何飞熊1, 谢海巍1, 蒲超2, 邹传铭2, 贾艺璇1. 基于改进YOLOv8网络的道路病害检测方法[J]. 计算机与现代化, 2025, 0(02): 108-113.
[13]	刘飞, 杨德刚, 章鑫, 秦静. 基于YOLOv8改进的水下目标检测算法[J]. 计算机与现代化, 2025, 0(01): 113-119.
[14]	谷岳, 邓松峰, 沈霁, 穆文涛, 赵恩棋. 基于改进YOLOv8的SAR舰船目标检测算法[J]. 计算机与现代化, 2024, 0(12): 78-83.
[15]	张思敏, 刘新妹, 殷俊龄, 李宝玲. 基于YOLOv7改进的PCB缺陷检测方法[J]. 计算机与现代化, 2024, 0(12): 45-52.