摘要
随着在线购物、自助结算等新零售商业模式的发展,运营终端对商品智能识别和结算的需求越来越高。人工智能在突破硬件限制以后得到突破性发展,其所用的卷积神经网络技术能很好地支撑商品图像识别应用。对卷积神经网络的系统架构和模型进行改进,尤其是对耗时较长的训练学习模型进行优化设计。经实际测试:该设计能有效提高商品图像识别系统计算的速度和识别准确性,满足社会在智能化商务运营中高效结算的业务需求。
With the development of new retail business models such as online shopping and self-service settlement,the demand for intelligent identification and settlement of goods is increasing.Artificial intelligence has made a breakthrough after breaking through the hardware limitations,and its convolution neural network technology can well support the application of commodity image recognition.The system architecture and model of convolutional neural network are improved,especially the time-consuming training and learning model is optimized.The actual test shows that it can effectively improve the calculation speed and recognition accuracy of commodity image recognition system,and meet the business needs of efficient settlement in intelligent business operation.
作者
蒋达央
JIANG Dayang(Changzhou College of Information Technology,Changzhou Jiangsu 213164,China)
出处
《北京工业职业技术学院学报》
2021年第3期28-31,共4页
Journal of Beijing Polytechnic College
关键词
卷积神经网络
商品图像识别
系统设计
图像处理
convolution neural network
commodity image recognition
system design
image processing
