首页|基于YOLOv8改进模型的管制刀具检测算法研究

基于YOLOv8改进模型的管制刀具检测算法研究

扫码查看
管制刀具虽然在日常生活中是必不可少的重要物品,但在安检中是禁止携带的,否则容易给犯罪分子以可乘之机,威胁社会安全.为了提高安检环节检测管制刀具效率,提出了基于YOLOv8 改进模型的检测算法.实验结果证明,在 YOLOv8 的基础上,通过在 Backbone 中添加CBAM注意力机制,并引入损失函数Wise-IoU,改进后的算法map提升了11%,达到60.4%,准确率提升5.6%,为后续识别管制刀具的算法提供了一定的参考价值.
Research on Control Tool Detection Algorithm Based on YOLOv8 Improved Model
Although controlled knives are essential and important items in daily life,they are prohibited from being carried during security checks,otherwise they may give criminals an opportunity to exploit and threaten social security.In order to improve the efficiency of tool detection and control in the security inspection process,this pa-per proposes a detection algorithm based on the YOLOv8 improved model.The experimental results show that,based on YOLOv8,by adding CBAM attention mechanism on Backbone and introducing the loss function Wise IoU,the improved algorithm map has been improved by 8.2%,reaching 60.4%,providing a certain reference value for subsequent algorithms for identifying controlled tools

YOLOv8CBAMloss function

齐万鹏

展开 >

天津职业技术师范大学,天津 300350

YOLOv8 CBAM 损失函数

2024

山西电子技术
山西省电子工业科学研究院 山西省电子学会

山西电子技术

影响因子:0.197
ISSN:1674-4578
年,卷(期):2024.(6)