摘要
将数据标注运用到智能家居系统中,通过摄像头、传感器等设备采集图片、视频、文本、语音数据,标注场景有面部识别、家具的标记、语音采集、人脸关键信息点提取、突发事件采集等.通过数据标注技术对人的行为、表情、言语进行识别,主要使用ImageNet数据集、Youtube-8M数据集、Yelp数据集、LibriSpeech数据集等分别作为图像、视频、文本和语音标注的标注数据数据集.对图像标注方法进行改善,通过图像标注对居民的安全处境、情绪状态等做出相应识别处理,从而解决家庭里的各种问题.
Abstract
The data labeling is applied to the smart home system.Through the camera,sensors and other devices,pictures,video,text,speech data are collected and labeled as facial recognition,home marking,speech collection,face key information point extraction,and emergency collection,etc.Through data labeling technology,we can identify people's behavior,expres-sions,speech,by mainly using ImageNet dataset,Youtube-8M dataset,Yelp dataset,LibriSpeech dataset as image annota-tion,video annotation,text annotation,speech annotation of the annotation data dataset.This paper improves the image anno-tation method,through the image annotation of the residents'security situation,emotional state,etc.It improves correspond-ing recognition processing,so as to solve various problems in the family.
基金项目
河北省物联网监控技术创新中心项目(21567693H)
青海省重点实验室项目(2017-ZJ-Y21)
中央高校基本科研业务费专项(3142021009)
廊坊市科技计划(2021011021)
维普
万方数据