基于深度学习的卷烟库存盘点图像识别技术研究及应用OA

This article proposes and implements a solution based on deep learning image recognition to address the core pain points of the tobacco industry's retail customers,such as time-consuming and inefficient manual inventory counting of cigarettes,high error rates,and negative impacts on customer experience.By constructing a multi-scenario cigarette image dataset and combining an improved YOLOv8 algorithm with lightweight model deployment technology,a"Retail Customer Cigarette Inventory Assistance Tool"mini-program has been developed.This technology not only improves the accuracy of cigarette recognition to 94.5%,but also achieves revolutionary improvements in customer service:it reduces the average time for retailers to complete a single inventory count from 2 minutes to less than 30 seconds,and the error rate is reduced to below 1%,significantly enhancing customer satisfaction and operational experience.At the same time,this tool provides tobacco company customer managers with an efficient digital service tool.Through automated inventory counting and data accumulation,it empowers customer managers to improve service efficiency and deepen operational guidance,achieving a transformation from traditional manual inspection to an intelligent and precise customer service model.This provides key technical support and a successful paradigm for the digital transformation of terminal services in the tobacco industry.