基于YOLOv5的卷烟包装真伪智能识别APP系统

YOLOv5-Based Intelligent Cigarette Packaging Authenticity Recognition Mobile APP System

在目前的卷烟市场稽查工作中,卷烟真伪的鉴别主要依赖于人眼感官,通过比对检验样品与真品以及真品的加工工艺和印刷工艺等方面的差异来判定,该方法存在效率低、主观性强的问题。为解决上述问题,提出了一种基于人工智能的卷烟包装真伪智能识别系统。该系统使用融合注意力机制的YOLOv5网络模型,用于识别卷烟包装图像的特征并判断其真伪;同时建立了卷烟包装图像信息的数据库和管理系统,开发了手机端应用APP软件,封装了识别算法,能实现卷烟包装的快速光学成像采集和图像真伪智能识别,以提高卷烟稽查工作的效率。

In the current cigarette market inspection work,the identification of cigarette authenticity mainly relies on human sensory perception.It involves determining the authenticity by comparing inspection samples with genuine products and examining differences in the genuine product′s processing and printing techniques.This method suffers from issues such as low efficiency and subjectivity.To address these problems,this paper proposes an AI-based intelligent identification system for cigarette packaging authenticity.This system utilizes the YOLOv5 network model with a fused attention mechanism to identify the features of cigarette packaging images and determine their authenticity.Simultaneously,a database and management system for cigarette packaging image information have been established,along with the development of a mobile application(APP)that encapsulates the recognition algorithm.This system enables rapid optical imaging acquisition of cigarette packaging and intelligent image authenticity recognition,thereby improving the efficiency of cigarette inspection work.