【摘要】 随着IT技术的迅速发展,图像识别技术已经慢慢融入并改变了人们的生活。为了使训练模型更加合理并帮助有特定图像识别需求的用户,此次研究采用卷积神经网络搭建一个可定制训练模型和服务的平台,并对2 762种不同品牌香烟进行测试分类识别。通过对不同品牌的香烟图片进行定制化训练,识别率平均可达96%,验证了该方案在实际应用中的有效性。更多还原

【Abstract】 With the rapid development of IT technology,image recognition technology has gradually integrated into and changed people’s lives.In order to make the training model more reasonable and help users with specific image recognition needs,this study uses convolutional neural network to build a platform for customized training model and services,and tests and classifies 2762 different brands of cigarettes.Through customized training for different brands of cigarette pictures,the recognition rate can reach 96% on average,which verifies the effectiveness of the scheme in practical application.更多还原