摘要
普通大众消费群体因中药材鉴定知识的缺乏,在购药时对药材的等级划分、真伪及优劣鉴定常陷于无具可用、无据可依的困境。为了解决这一困境,以西洋参为例,基于百度人工智能开放平台提供的EasyDL定制化图像识别服务,以安卓手机为前端、云服器为后台设计了一种基于图像识别技术的中药材品鉴助手系统,安卓手机前端软件采用微信小程序架构实现图像的采集与上传,后台云服器调用百度AI图像识别服务实现了中药材的等级分析。实验测试结果显示,在白纸背景下拍照分析的识别正确率高达95%,能满足实际使用的要求。
In view of the fact that ordinary consumers lack knowledge of Chinese medicine and can not accurately identify Chinese medicinal materials when purchasing them, taking Panax quinquefolium for example, a Chinese herbal medicine recognition system was designed based on EasyDL customized image recognition service provided by Baidu artificial intelligence open platform. The system is made up of Android mobile front end and cloud server background. The Android mobile front-end software uses WeChat small program architecture. The Android front-end is used for image acquisition and uploading. The background cloud service calls Baidu AI image recognition service to realize the grade analysis of Chinese medicinal materials. The experimental results show that the recognition accuracy of photographic analysis in white paper background is as high as 95%, which can meet the requirements of practical use.
引文
[1]张喜红.基于BP神经网络的西洋参等级分类方法研究[J].云南民族大学学报(自然科学版),2017(4):322-326.
[2]张喜红,王玉香.基于Scrapy的中药材网络信息采集方法研究[J].江汉大学学报(自然科学版),2018(6):522-527.
[3]连自锋.基于深层神经网络的图像识别算法研究[D].北京:北京邮电大学,2017.
[4]王会,于栋祯.眩晕疾病人工智能专家诊疗系统研究进展[J].第二军医大学学报,2018(8):935-938.
[5]黄欢,赵钢.人工智能在医疗及神经病学领域的应用[J].华西医学,2018(6):639-643.
[6]钟义信.人工智能:“热闹”背后的“门道”[J].科技导报,2017(6):14-19.
[7]李成吉,张淑娟.计算机视觉技术在核桃壳、仁及分心木识别中的应用研究[J].山西农业大学学报(自然科学版),2018(11):20-26.
[8]傅向葵,傅一迪.基于人工智能非金属夹杂物的检测与研究[J].中国冶金,2014(11):14-18.
[9]马述清.PHP网络编程[M].北京:电子工业出版社,2014.