基于yolov3的水果蔬菜检测算法研究  

Research on fruit and vegetable detection algorithm based on yolov3

作　　者：范杰 郭原东 李向阳 张骁 王建云 FAN Jie;GUO Yuandong;LI Xiangyang;ZHANG Xiao;WANG Jianyun(Hubei Key Laboratory of Intelligent Vision based Monitoring for Hydroelectric Engineering,Three Gorges University,YiChang 443002,China;College of Computer and Information Technology,Three Gorges University,YiChang 443002,China)

机构地区：[1]三峡大学水电工程智能视觉监测湖北省重点实验室,湖北宜昌443002 [2]三峡大学计算机与信息学院,湖北宜昌443002

出　　处：《长江信息通信》2022年第1期3-6,共4页Changjiang Information & Communications

摘　　要：为减轻超市等需人工称重售卖体系在人流高峰期的售卖压力,同时响应当前国家对疫情防控中减少公共场所人员接触防疫政策,实现超市中水果蔬菜的智能化售卖也是当前实现智慧城市建设的一部分。其中智能化售卖中最关键的问题是实现对水果蔬菜的精确识别;文章采用YOLOv3网络算法模型,在自己采集构建的水果蔬菜数据集上进行训练和测试,并与YOLOv3-Tiny网络进行对比实验。实验结果表明,YOLOv3在检测准确率上表现得更好,检测精度达到了94.69%,而轻量型网络YOLOv3-Tiny则在检测的效率上表现得更好。In order to alleviate the sales pressure of supermarkets and other manual weighing sales systems during the peak of the flow of people, and at the same time respond to the current national epidemic prevention and control policies to reduce the exposure of people in public places to the epidemic, the realization of intelligent sales of fruits and vegetables in supermarkets is also the current realization of the construction of smart cities. a part of. Among them, the most critical problem in intelligent sales is to realize the accurate identification of fruits and vegetables;this article uses the YOLOv3 network algorithm model to train and test on the fruit and vegetable data set collected and constructed by myself, and conduct a comparative experiment with the YOLOv3-Tiny network. Experimental results show that YOLOv3 performs better in detection accuracy, with a detection accuracy of 94.69%, while the lightweight network YOLOv3-Tiny performs better in detection efficiency.

关 键 词：水果蔬菜检测 YOLOv3 YOLOv3-Tiny 检测精度 检测效率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]