摘要
研究并设计了一个面向短视频不良内容的实时检测平台。该平台研究的核心在于分层筛选系统,通过基于短视频外围核心参数构建的深度学习筛选模型完成海量筛选,再将检测出的可疑不良短视频传递给基于深度学习的短视频内容识别引擎进行重点甄别,提出融合自然语言处理、计算机视觉、语音识别、机器学习等的短视频智能实时检测平台框架。
A platform for real-time short video bad content detecting is designed. The core of the research on the platform is a hierarchical screening system. A large-scale screening is performed through a deep learning screening model constructed based on the peripheral core parameters of short video, and the detected suspicious bad short videos are passed to the deep learning based short video content recognition engine for key screening. Thus a short video intelligent real-time detection platform framework,which incorporates natural language processing, computer vision, speech recognition, and machine learning, is put forward.
作者
周利红
张军军
常孔帅
Zhou Lihong;Zhang Junjun;Chang Kongshuai(Zhejiang Radio and TV University,Hangzhou,Zhejiang 310012,China;Zhejiang Jolly Information Technology Co.Ltd;Zhejiang University of Technology)
出处
《计算机时代》
2020年第5期38-40,共3页
Computer Era
关键词
短视频
不良内容
智能检测
分层筛选
short video
bad content
intelligent detection
hierarchical screening