YOLOv8n-HCDP：轻量化木材缺陷检测模型

唐健强; 徐梓敬; 徐凯宏; 程仁轩; 高俊哲

doi:10.12326/j.2096-9694.2025001

研究报告 | 浏览量 : 0 下载量: CSCD: 0 CNKI被引量: 0 Crossref: 0 Wos: 0 Scopus: 0 更多指标

PDF
导出
分享
收藏
专辑

YOLOv8n-HCDP：轻量化木材缺陷检测模型
YOLOv8n-HCDP: Lightweight Wood Defect Detection Model
1.东北林业大学计算机与控制工程学院，黑龙江哈尔滨 150040
2.东北林业大学家居与艺术设计学院，黑龙江哈尔滨 150040
2025年39卷第4期页码：89-97
DOI： 10.12326/j.2096-9694.2025001
稿件说明：

移动端阅览

唐健强,徐梓敬,徐凯宏等.YOLOv8n-HCDP：轻量化木材缺陷检测模型[J].木材科学与技术,2025,39(04):89-97. DOI： 10.12326/j.2096-9694.2025001.

TANG Jianqiang,XU Zijing,XU Kaihong,et al.YOLOv8n-HCDP: Lightweight Wood Defect Detection Model[J].Chinese Journal of Wood Science and Technology,2025,39(04):89-97. DOI： 10.12326/j.2096-9694.2025001.

摘要

针对木材缺陷检测领域中深度学习模型参数量大，分类检测准确率不高的问题，提出一种基于YOLOv8n的轻量化检测模型YOLOv8n-HCDP。首先，构建轻量化骨干网络HgNetv2（high performance GPU network v2），随后利用动态头（dynamic head）融合轻量级跨尺度特征融合模块（cross-scale feature fusion module，CCFM）得到全新的CCFM-dy模块代替传统的颈部网络和检测头，从而减少模型参数量和计算量；引入动态卷积（dynamic convolution），使网络在保持低计算量的同时从大规模训练中获益；最后，引入创新的PPC结构替换C2f（CSP bottleneck）结构，进一步轻量化模型。结果表明：相较于基准模型，改进模型的参数量减少54.15%，计算量减少44.44%，体积减少51.42%，平均精度均值（mean average precision，mAP）提升2.0%，更适合在硬件资源受限的嵌入式设备上部署。

Abstract

Aiming at the problems of large number of deep learning model parameters and low classification and detection accuracy in the field of wood defect detection

a lightweight detection model YOLOv8N-HCDP based on YOLOv8n was proposed. Firstly

the lightweight backbone network of HgNetv2 (high performance GPU network v2) is constructed. Secondly

a new CCFM-dy module is obtained by Dynamic Head fusion with lightweight cross-scale feature fusion module (CCFM) to replace the traditional neck network and detection head

reducing the number of model parameters and calculation amount. Dynamic convolution is introduced to make the network benefit from large-scale training while maintaining low computation. Finally

an innovative PPC structure is introduced to replace CSP bottleneck(C2f)in the network structure to further lightweight the model. The experimental results show that compared with the benchmark model

the improved model has 54.15% less parameters

44.44% less computation

51.42% less volume

and 2.0% more mAP50

which is more suitable for deployment on embedded devices with limited hardware resources. It provides a more efficient defect detection solution for the wood processing industry.

关键词

Keywords

references

KILIÇ K , KILIÇ K , DOĞRU İ A , et al . WD Detector: deep learning-based hybrid sensor design for wood defect detection [J ] . European Journal of Wood and Wood Products , 2025 , 83 ( 2 ): 1 - 13 .

贺春光 , 李璐芳 , 高峰 , 等 . 基于核主成分分析的GSA-SVM木材单板缺陷识别研究 [J ] . 森林工程 , 2023 , 39 ( 2 ): 91 - 99 .

HE C G , LI L F , GAO F , et al . Study on GSA-SVM wood veneer defect identification based on kernel principal component analysis [J ] . Forest Engineering , 2023 , 39 ( 2 ): 91 - 99 .

YU H L , LIANG Y L , LIANG H , et al . Recognition of wood surface defects with near infrared spectroscopy and machine vision [J ] . Journal of Forestry Research , 2019 , 30 ( 6 ): 2379 - 2386 .

WANG R J , LIANG F L , WANG B , et al . ODCA-YOLO: an omni-dynamic convolution coordinate attention-based YOLO for wood defect detection [J ] . Forests , 2023 , 14 ( 9 ): 1885 .

朱豪 , 周顺勇 , 曾雅兰 , 等 . 基于改进YOLOv5s的木材表面缺陷检测模型 [J ] . 木材科学与技术 , 2023 , 37 ( 2 ): 8 - 15 .

ZHU H , ZHOU S Y , ZENG Y L , et al . Detection model of wood surface defects based on improved YOLOv5s [J ] . Chinese Journal of Wood Science and Technology , 2023 , 37 ( 2 ): 8 - 15 .

AN H , LIANG Z H , QIN M M , et al . Wood defect detection based on the CWB-YOLOv8 algorithm [J ] . Journal of Wood Science , 2024 , 70 ( 1 ): 26 .

SONG B Y , CHEN JY , LIU W Bet al . YOLO-SM: A lightweight single-class multi-deformation object detection network [J ] . IEEE Transactions on Emerging Topics in Computational Intelligence , 2024 , 8 ( 3 ): 2467 - 2480 .

YANG R-X , LEE Y-R , LEE F S , et al . Improvement of YOLO detection strategy for detailed defects in bamboo strips [J ] . Forests , 2025 , 16 ( 4 ): 595 .

XI H L , WANG R J , LIANG F L , et al . SiM-YOLO: a wood surface defect detection method based on the improved YOLOv8 [J ] . Coatings , 2024 , 14 ( 8 ): 1001 .

王文财 , 党亚光 , 朱翔 , 等 . 基于Lite-YOLOv5s模型的刨花板表面缺陷检测方法 [J ] . 木材科学与技术 , 2023 , 37 ( 3 ): 58 - 67 .

WANG W C , DANG Y G , ZHU X , et al . Detecting method for particleboard surface defects based on the lite-YOLOv5s model [J ] . Chinese Journal of Wood Science and Technology , 2023 , 37 ( 3 ): 58 - 67 .

ZENG T H , LI S Y , SONG Q M , et al . Lightweight tomato real-time detection method based on improved YOLO and mobile deployment [J ] . Computers and Electronics in Agriculture , 2023 , 205 : 107625 .

LI S L , SUN S J , LIU Y , et al . Real-time lightweight YOLO model for grouting defect detection in external post-tensioned ducts via infrared thermography [J ] . Automation in Construction , 2024 , 168 : 105830 .

NIMMA D , AL-OMARI O , PRADHAN R , et al . Object detection in real-time video surveillance using attention based transformer-YOLOv8 model [J ] . Alexandria Engineering Journal , 2025 , 118 : 482 - 495 .

MA B L , HUA Z X , WEN Y C , et al . Using an improved lightweight YOLOv8 model for real-time detection of multi-stage apple fruit in complex orchard environments [J ] . Artificial Intelligence in Agriculture , 2024 , 11 : 70 - 82 .

YASIR M , LIU S W , Pirasteh S , et al . YOLOShipTracker: Tracking ships in SAR images using lightweight YOLOv8 [J ] . International Journal of Applied Earth Observation and Geoinformation , 2024 , 134 : 104137 .

孙嘉傲 . 基于改进YOLO与DETR的X光违禁品检测研究 [D ] . 北京：中国人民公安大学 , 2024 .

ZHAO Y A , LV W Y , XU S L , et al . DETRs beat YOLOs on real-time object detection [C ] // 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . June 16-22, 2024 , Seattle, WA, USA. IEEE , 2024 : 16965 - 16974 .

LIAO H J , WANG G P , Jin S Y , et al . HCRP-YOLO: a lightweight algorithm for potato defect detection [J ] . Smart Agricultural Technology , 2025 , 10 : 100849 .

DAI X Y , CHEN Y P , XIAO B , et al . Dynamic head: unifying object detection heads with attentions [C ] // 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . June 20-25, 2021 , Nashville, TN, USA. IEEE , 2021 : 7369 - 7378 .

HAN K , WANG Y H , GUO J Y , et al . ParameterNet: parameters are all you need for large-scale visual pretraining of mobile networks [C ] // 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . June 16-22, 2024 , Seattle, WA, USA. IEEE , 2024 : 15751 - 15761 .

CHEN J R , KAO S H , HE H , et al . Run, don’t walk: chasing higher FLOPS for faster neural networks [C ] // 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . June 17-24, 2023 , Vancouver, BC, Canada . IEEE , 2023 : 12021 - 12031 .

田朔 , 刘美怡 , 杨东 , 等 . 基于木材纹理图像和改进ResNet50＿DTPE模型的5种红木树种识别方法 [J ] . 木材科学与技术 , 2024 , 38 ( 5 ): 77 - 84 .

TIAN S , LIU M Y , YANG D , et al . Identification method for five hongmu species based on wood texture images and an improved ResNet50＿DTPE model [J ] . Chinese Journal of Wood Science and Technology , 2024 , 38 ( 5 ): 77 - 84 .

CARION N , MASSA F , SYNNAEVE G , et al . End-to-end object detection with transformers [C ] // Computer Vision – ECCV 2020 . Cham : Springer International Publishing , 2020 : 213 - 229 .

文章被引用时，请邮件提醒。

提交

基于改进YOLOv10s的木材表面缺陷检测模型

基于改进LeNet-5模型的木材表面典型缺陷识别方法研究