基于ShuffleNetV2与CSPPC的YOLOv8n铝型材缺陷检测轻量化模型

打开文本图片集

中图分类号：TP391.4 文献标识码：A

文章编号：1006-8228（2025）11-32-04

LightweightModel forDefect Detection ofYOLOv8n Aluminum Profiles Based on ShuffleNetV2 and CSPPC

Shen Jiamin

（Shanxi EngineeringVocational College，Taiyuan，Shanxi O3ooo9，China）

Abstract：Thisarticlefocusesontheproblemsoflargediferencesindefectscale，complexbackgroundinterference，andeasy misseddetectionofsmalldefectsinthesurfacedefectdetectiontaskofaluminumprofiles，andtomeetlowcomputationalcost requirements，thisarticlehasimprovedtheYOLOv8nmodelFirstlywereplacethebackbonenetworkwiththelightweight ShufleetV2networktoreducethenumberofmodelparameters；SecondlyweintroducetheCSPPCmoduleinthefeaturefusion section to reduce computational redundancy through partial convolution.Theexperimental resultsshow that the mAP@50 of the improved model reaches 81.3% ，an increase of 3 percentage points compared to the original model，a reduction of 21.6% in parameter count，and a reduction of 19.5% incomputational complexity. It reduces parameter count and computational complexity while improving detection accuracy.

Keywords：Defect Detection;YOLOv8n;Lightweight;ShuffleNetV2

0引言

铝型材广泛应用于制造业和建筑业，其表面质量直接影响产品性能。（剩余5472字）