人工智能驱动下的图像识别算法优化与应用

——以某特大型交通枢纽安检场景为例

打开文本图片集

doi：10.3969/J.ISSN.1672-7274.2025.06.035

中图分类号：TP18；TP391.41；U495 文献标志码：B 文章编码：1672-7274（2025）06-0104-03

Optimization and Application of Image Recognition Algorithms Driven byArtificial Intelligence

-Taking the Security Check Scene of a Certain Mega Transportation Hub as an Example GUANLingling1，GUANLingting² （1.Mount Taishan Vocational and Technical College，Tai'an 271ooo，China; 2. Shandong Lubei Construction Engineering Co.，Ltd.， Dezhou 253ooo， China）

Abstract： Image recognition technology based on deep learning faces high concurrency and high accuracy requirements in the applicationof large-scale transportation hub security check scenarios.The article proposes a dangerous goods recognition algorithm basedon improvedYOLOv5，which integrates crosslayer feature enhancement module and channel atention mechanism，to address the security check demand of 3oooo daily passengers at a certain mega transportation hub.The background of the entire article needs to be modified.The algorithm can be optimizedandsimulated for application，but not fora specific project.Through laboratory simulation testing and datavalidation， the optimized algorithmachieved an accuracy rate of 96.3% inidentifying prohibited items such as controlled knives and flammable liquids，which is 41.2% higher than traditional manual inspections. The daily security processing capacity has also increased by 28.6% .Thisalgorithmscheme provides new technical ideas for the development of intelligent security check systems in large-scale transportation hubs.

Keywords： securitycheck system; object detection; YOLOv5; atention mechanism; identification of dangerous goods

大型交通枢纽人流密集，安检压力与日俱增。（剩余4229字）