基于改进YOLOv7的电梯内电动车识别

打开文本图片集

中图分类号：TP391 文献标志码：A 文章编号：1671-5276（2025）06-0136-05

Abstract：Asfortheloweficiency，pooraccuracyandhighlaborcostinidentifying electricvehicles inelevators，animproved method basedonYouOnlyLookOnceversion7wasproposed.GlobalAtention Mechanismwasintorducedintothehead network，and ELAN module wasrepalcedbyareparamererized eficient feature extraction modulenamedRepGhostNeXtin backbone network，reducing resource consumption and enhancing detection accuracy.The results show that the of the improvedmodel reaches 89.9% ，4.2percentages higher than the original YOLOv7，the recognition accuracy of the electric vehicles isas highas97.9%，andtheGFlopsof the model isreduced by97.4G，whichhas high applicationvalueinidentifying electric vehiclesin elevators.

Keywords：e-vehicles detection；YOLOv7；target recognition；GAM；RepGhostNeXt

0 引言

电梯作为常见的垂直交通工具，被广泛应用于住宅小区、办公楼、商场等人员密集场所，为人们出行带来了极大的便利。（剩余6361字）