《计算机应用》唯一官方网站 ›› 2025, Vol. 45 ›› Issue (2): 647-654.DOI: 10.11772/j.issn.1001-9081.2024020198
• 多媒体计算与计算机仿真 • 上一篇
收稿日期:
2024-02-28
修回日期:
2024-06-08
接受日期:
2024-06-17
发布日期:
2024-07-19
出版日期:
2025-02-10
通讯作者:
余松森
作者简介:
林智凡(2001—),男,广东汕尾人,硕士研究生,主要研究方向:计算机视觉基金资助:
Songsen YU(), Zhifan LIN, Guopeng XUE, Jianyu XU
Received:
2024-02-28
Revised:
2024-06-08
Accepted:
2024-06-17
Online:
2024-07-19
Published:
2025-02-10
Contact:
Songsen YU
About author:
LIN Zhifan, born in 2001, M. S. candidate. His research interests include computer vision.Supported by:
摘要:
针对当前瓷砖缺陷检测主要依靠人工检测导致的主观性强、效率低、劳动强度大等问题,提出一种基于改进YOLOv8的轻量级大幅面瓷砖图像微小缺陷检测算法。首先,对高分辨率大幅面图像进行裁切处理,并在骨干网络中引入HorBlock增强模型的捕捉能力;其次,融入大型可分离内核注意力(LSKA)改进C2f提高模型的检测性能,并通过引入SA(Shuffle Attention)增强模型的特征提取能力;最后,引入全维度动态卷积(ODConv)进一步增强模型对微小缺陷的处理能力。在阿里天池瓷砖瑕疵检测数据集上的实验结果表明:改进后的模型不仅参数量比原始YOLOv8n低,而且mAP@0.5提升了8.2个百分点,F1分数提升了7个百分点。可见,改进后的模型能更精确地识别和处理大幅面瓷砖的微小表面缺陷,且能在保持轻量级的同时,显著提升检测效果。
中图分类号:
余松森, 林智凡, 薛国鹏, 徐建宇. 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.
Songsen YU, Zhifan LIN, Guopeng XUE, Jianyu XU. Lightweight large-format tile defect detection algorithm based on improved YOLOv8[J]. Journal of Computer Applications, 2025, 45(2): 647-654.
缺陷类别 | mAP@0.5 | mAP@0.5:0.95 |
---|---|---|
所有 | 74.5 | 42.7 |
边异常(edge) | 91.9 | 59.5 |
角异常(corner) | 94.5 | 64.0 |
白色点(white) | 53.7 | 21.2 |
浅色块(gray) | 56.7 | 27.7 |
深色点(black) | 66.9 | 29.7 |
光圈(circle) | 83.2 | 54.2 |
表1 切片后的mAP值 (%)
Tab. 1 mAP value after slicing
缺陷类别 | mAP@0.5 | mAP@0.5:0.95 |
---|---|---|
所有 | 74.5 | 42.7 |
边异常(edge) | 91.9 | 59.5 |
角异常(corner) | 94.5 | 64.0 |
白色点(white) | 53.7 | 21.2 |
浅色块(gray) | 56.7 | 27.7 |
深色点(black) | 66.9 | 29.7 |
光圈(circle) | 83.2 | 54.2 |
模型 | mAP@0.5/% | map@0.5:0.95/% | Params/106 | GFLOPs | F1/% | FPS |
---|---|---|---|---|---|---|
YOLOv8n | 74.5 | 42.7 | 3.0 | 8.1 | 71 | 435 |
+HorBLock | 79.0 | 45.1 | 3.1 | 9.2 | 75 | 286 |
+HorBLock+LSKA | 80.1 | 45.2 | 2.8 | 8.9 | 76 | 263 |
+HorBLock+LSKA+SA | 81.9 | 45.9 | 2.8 | 8.9 | 78 | 278 |
+HorBLock+LSKA+SA+ODConv | 82.7 | 46.8 | 2.8 | 8.5 | 78 | 270 |
表2 消融实验结果
Tab. 2 Ablation experiment results
模型 | mAP@0.5/% | map@0.5:0.95/% | Params/106 | GFLOPs | F1/% | FPS |
---|---|---|---|---|---|---|
YOLOv8n | 74.5 | 42.7 | 3.0 | 8.1 | 71 | 435 |
+HorBLock | 79.0 | 45.1 | 3.1 | 9.2 | 75 | 286 |
+HorBLock+LSKA | 80.1 | 45.2 | 2.8 | 8.9 | 76 | 263 |
+HorBLock+LSKA+SA | 81.9 | 45.9 | 2.8 | 8.9 | 78 | 278 |
+HorBLock+LSKA+SA+ODConv | 82.7 | 46.8 | 2.8 | 8.5 | 78 | 270 |
模型 | mAP@0.5/% | map@0.5:0.95/% | Params/106 | GFLOPs | F1/% | FPS |
---|---|---|---|---|---|---|
YOLOv3-tiny | 74.6 | 42.3 | 12.1 | 18.9 | 71 | 455 |
YOLOv5n | 76.5 | 40.4 | 2.5 | 7.1 | 72 | 455 |
YOLOv5s | 81.5 | 45.6 | 9.1 | 23.8 | 77 | 323 |
YOLOv6n | 65.7 | 36.2 | 4.2 | 11.8 | 64 | 476 |
YOLOv7-tiny | 39.1 | 14.0 | 6.0 | 13.1 | 46 | 161 |
YOLOv8s | 83.5 | 49.3 | 11.1 | 28.4 | 79 | 323 |
YOLOv8-MobileNetV3 | 72.8 | 37.5 | 2.5 | 5.9 | 69 | 417 |
YOLOv8-VanillaNet | 68.0 | 36.9 | 2.0 | 5.9 | 65 | 476 |
DAMO-YOLO-T | 79.4 | 44.7 | 8.2 | 17.2 | — | 182 |
Gold-YOLOn | 69.2 | 36.9 | 5.6 | 12.1 | 68 | 370 |
Tile-YOLO | 82.7 | 46.8 | 2.8 | 8.5 | 78 | 270 |
表3 不同模型的实验结果对比
Tab. 3 Comparison of experimental results of different models
模型 | mAP@0.5/% | map@0.5:0.95/% | Params/106 | GFLOPs | F1/% | FPS |
---|---|---|---|---|---|---|
YOLOv3-tiny | 74.6 | 42.3 | 12.1 | 18.9 | 71 | 455 |
YOLOv5n | 76.5 | 40.4 | 2.5 | 7.1 | 72 | 455 |
YOLOv5s | 81.5 | 45.6 | 9.1 | 23.8 | 77 | 323 |
YOLOv6n | 65.7 | 36.2 | 4.2 | 11.8 | 64 | 476 |
YOLOv7-tiny | 39.1 | 14.0 | 6.0 | 13.1 | 46 | 161 |
YOLOv8s | 83.5 | 49.3 | 11.1 | 28.4 | 79 | 323 |
YOLOv8-MobileNetV3 | 72.8 | 37.5 | 2.5 | 5.9 | 69 | 417 |
YOLOv8-VanillaNet | 68.0 | 36.9 | 2.0 | 5.9 | 65 | 476 |
DAMO-YOLO-T | 79.4 | 44.7 | 8.2 | 17.2 | — | 182 |
Gold-YOLOn | 69.2 | 36.9 | 5.6 | 12.1 | 68 | 370 |
Tile-YOLO | 82.7 | 46.8 | 2.8 | 8.5 | 78 | 270 |
缺陷 类别 | YOLOv8n | Tile-YOLO | ||
---|---|---|---|---|
mAP@0.5 | mAP@0.5:0.95 | mAP@0.5 | mAP@0.5:0.95 | |
所有 | 74.5 | 42.7 | 82.7 | 46.8 |
边异常 | 91.9 | 59.5 | 88.0 | 54.5 |
角异常 | 94.5 | 64.0 | 94.7 | 63.7 |
白色点 | 53.7 | 21.2 | 75.7 | 30.4 |
浅色块 | 56.7 | 27.7 | 68.2 | 35.1 |
深色点 | 66.9 | 29.7 | 82.6 | 38.1 |
光圈 | 83.2 | 54.2 | 87.1 | 59.4 |
表4 不同模型对各类缺陷的识别性能 (%)
Tab. 4 Detecting performance of various defects by different models
缺陷 类别 | YOLOv8n | Tile-YOLO | ||
---|---|---|---|---|
mAP@0.5 | mAP@0.5:0.95 | mAP@0.5 | mAP@0.5:0.95 | |
所有 | 74.5 | 42.7 | 82.7 | 46.8 |
边异常 | 91.9 | 59.5 | 88.0 | 54.5 |
角异常 | 94.5 | 64.0 | 94.7 | 63.7 |
白色点 | 53.7 | 21.2 | 75.7 | 30.4 |
浅色块 | 56.7 | 27.7 | 68.2 | 35.1 |
深色点 | 66.9 | 29.7 | 82.6 | 38.1 |
光圈 | 83.2 | 54.2 | 87.1 | 59.4 |
1 | HANZAEI S H, AFSHAR A, BARAZANDEH F. Automatic detection and classification of the ceramic tiles’ surface defects[J]. Pattern Recognition, 2017, 66: 174-189. |
2 | MATIĆ 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. |
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. |
LI J H, QUAN X X, WANG Y L. Research on defect detection algorithm of ceramic tile surface with multi-feature fusion[J]. Computer Engineering and Applications, 2020, 56(15): 191-198. | |
4 | DONG G, SUN S, WU N, et al. A rapid detection method for the surface defects of mosaic ceramic tiles[J]. Ceramics International, 2022, 48(11): 15462-15469. |
5 | HU H, ZHU Z. Sim-YOLOv5s: a method for detecting defects on the end face of lithium battery steel shells[J]. Advanced Engineering Informatics, 2023, 55: No.101824. |
6 | PARLAK İ E, EMEL E. Deep learning-based detection of aluminum casting defects and their types[J]. Engineering Applications of Artificial Intelligence, 2023, 118: No.105636. |
7 | USTABAŞ KAYA G. Development of hybrid optical sensor based on deep learning to detect and classify the micro-size defects in printed circuit board[J]. Measurement, 2023, 206: No.112247. |
8 | REN S, HE K, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[C]// Proceedings of the 28th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2015: 91-99. |
9 | LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]// Proceedings of the 2016 European Conference on Computer Vision, LNCS 9905. Cham: Springer, 2016: 21-37. |
10 | LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2017: 2999-3007. |
11 | CAO T, SONG K, XU L, et al. Balanced multi-scale target score network for ceramic tile surface defect detection[J]. Measurement, 2024, 224: No.113914. |
12 | 余松森,张明威,杨欢. 改进YOLOv5的瓷砖表面缺陷检测[J]. 计算机系统应用, 2023, 32(8):151-161. |
YU S S, ZHANG M W, YANG H. Detection of ceramic tile surface defects based on improved YOLOv5[J]. Computer Systems & Applications, 2023, 32(8): 151-161. | |
13 | CAO Y, PANG D, ZHAO Q, et al. Improved YOLOv8-GD deep learning model for defect detection in electroluminescence images of solar photovoltaic modules[J]. Engineering Applications of Artificial Intelligence, 2024, 131: No.107866. |
14 | LI H, LI J, WEI H, et al. Slim-neck by GSConv: a better design paradigm of detector architectures for autonomous vehicles[EB/OL]. [2023-12-19].. |
15 | TAN M, PANG R, 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. |
16 | CHEN S, LI Y, ZHANG Y, et al. Soft X-ray image recognition and classification of maize seed cracks based on image enhancement and optimized YOLOv8 model[J]. Computers and Electronics in Agriculture, 2024, 216: No.108475. |
17 | LIU L, LI P, WANG D, et al. A wind turbine damage detection algorithm designed based on YOLOv8[J]. Applied Soft Computing, 2024, 154: No.111364. |
18 | XIONG C, ZAYED T, MOHAMMED ABDELKADER E. A novel YOLOv8-GAM-Wise-IoU model for automated detection of bridge surface cracks[J]. Construction and Building Materials, 2024, 414: No.135025. |
19 | RAO Y, ZHAO W, TANG Y, et al. HorNet: efficient high-order spatial interactions with recursive gated convolutions[C]// Proceedings of the 36th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2022: 10353-10366. |
20 | LAU K W, PO L M, REHMAN Y A UR. Large separable kernel attention: rethinking the large kernel attention design in CNN[J]. Expert Systems with Applications, 2024, 236: No.121352. |
21 | ZHANG Q L, YANG Y B. SA-Net: shuffle attention for deep convolutional neural networks[C]// Proceedings of the 2021 IEEE International Conference on Acoustics, Speech and Signal Processing. Piscataway: IEEE, 2021: 2235-2239. |
22 | LI C, ZHOU A, YAO A. Omni-dimensional dynamic convolution[EB/OL]. [2023-12-19].. |
23 | 阿里云天池. 瓷砖瑕疵检测数据集[DS/OL]. [2023-12-15].. |
Tianchi Aliyun. Tile defect detection data set[DS/OL]. [2023-12- 15].. | |
24 | YANG B, BENDER G, LE Q V, et al. CondConv: conditionally parameterized convolutions for efficient inference[C]// Proceedings of the 33rd International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2019: 1307-1318. |
25 | CHEN Y, DAI X, LIU M, et al. Dynamic convolution: attention over convolution kernels[C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 11027-11036. |
26 | XU X, JIANG Y, CHEN W, et al. DAMO-YOLO: a report on real-time object detection design[EB/OL]. [2023-12-19].. |
27 | WANG C, HE W, NIE Y, et al. Gold-YOLO: efficient object detector via gather-and-distribute mechanism[C]// Proceedings of the 37th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2023: 51094-51112. |
28 | HOWARD A, SANDLER M, CHU G, et al. Searching for MobileNetV3[C]// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2019: 1314-1324. |
29 | CHEN H, WANG Y, GUO J, et al. VanillaNet: the power of minimalism in deep learning[C]// Proceedings of the 37th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2023: 7050-7064. |
[1] | 洪梓榕, 包广清. 基于集成学习的雷达自动目标识别综述[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 371-382. |
[2] | 张众维, 王俊, 刘树东, 王志恒. 多尺度特征融合与加权框融合的遥感图像目标检测[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 633-639. |
[3] | 孟海腾, 赵小乐, 李天瑞. 基于非对称信息蒸馏网络的轻量级图像超分辨重建[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 601-609. |
[4] | 邓淼磊, 阚雨培, 孙川川, 徐海航, 樊少珺, 周鑫. 基于深度学习的网络入侵检测系统综述[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 453-466. |
[5] | 何秋润, 胡节, 彭博, 李天源. 基于上下文信息的多尺度特征融合织物疵点检测算法[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 640-646. |
[6] | 张天骐, 谭霜, 沈夕文, 唐娟. 融合注意力机制和多尺度特征的图像水印方法[J]. 《计算机应用》唯一官方网站, 2025, 45(2): 616-623. |
[7] | 陈俊颖, 郭士杰, 陈玲玲. 基于解耦注意力与幻影卷积的轻量级人体姿态估计[J]. 《计算机应用》唯一官方网站, 2025, 45(1): 223-233. |
[8] | 刘赏, 周煜炜, 代娆, 董林芳, 刘猛. 融合注意力和上下文信息的遥感图像小目标检测算法[J]. 《计算机应用》唯一官方网站, 2025, 45(1): 292-300. |
[9] | 张思齐, 张金俊, 王天一, 秦小林. 基于信号时态逻辑的深度时序事件检测算法[J]. 《计算机应用》唯一官方网站, 2025, 45(1): 90-97. |
[10] | 郑宗生, 杜嘉, 成雨荷, 赵泽骋, 张月维, 王绪龙. 用于红外-可见光图像分类的跨模态双流交替交互网络[J]. 《计算机应用》唯一官方网站, 2025, 45(1): 275-283. |
[11] | 徐欣然, 张绍兵, 成苗, 张洋, 曾尚. 基于多路层次化混合专家模型的轴承故障诊断方法[J]. 《计算机应用》唯一官方网站, 2025, 45(1): 59-68. |
[12] | 梁杰涛, 罗兵, 付兰慧, 常青玲, 李楠楠, 易宁波, 冯其, 何鑫, 邓辅秦. 基于坐标几何采样的点云配准方法[J]. 《计算机应用》唯一官方网站, 2025, 45(1): 214-222. |
[13] | 晏燕, 钱星颖, 闫鹏斌, 杨杰. 位置大数据的联邦学习统计预测与差分隐私保护方法[J]. 《计算机应用》唯一官方网站, 2025, 45(1): 127-135. |
[14] | 秦璟, 秦志光, 李发礼, 彭悦恒. 基于概率稀疏自注意力神经网络的重性抑郁疾患诊断[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2970-2974. |
[15] | 王熙源, 张战成, 徐少康, 张宝成, 罗晓清, 胡伏原. 面向手术导航3D/2D配准的无监督跨域迁移网络[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2911-2918. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||