基于内容的视频高性能处理框架设计  

Design of High Performance Video Processing Framework Based on Content

作　　者：王惠峰 张峰 张昆 王子玮 白立飞 葛建军 张德 WANG Huifeng;ZHANG Feng;ZHANG Kun;WANG Ziwei;BAI Lifei;GE Jianjun;ZHANG De(Key Laboratory of Cognitive and Intelligent Technology,Information Science Academy of China Electronic Science and Technology Group Corporation,Beijing 100086,China)

机构地区：[1]中国电子科技集团公司信息科学研究院认知与智能技术重点实验室,北京100086

出　　处：《指挥信息系统与技术》2022年第2期85-90,共6页Command Information System and Technology

摘　　要：视频内容理解技术是实现视频内容监管、推荐和追踪的主要手段。为了实现海量和实时视频数据的快速处理,提出了一种基于消息总线及容器虚拟化技术的视频内容理解并发处理框架。通过分布式消息总线完成了视频处理数据/任务的缓存与分发,确保了视频处理过程的平稳和可靠运行。采用任务并行和模型并行2种机制实现了视频内容理解功能的并发处理,提高了视频处理效率,同时支持视频内容人工智能(AI)处理功能模块的无缝接入及大规模部署。Video content understanding technology is the main means to realize video content supervision,recommendation and tracking.To realize the massive and real-time video data rapidly processing,a video content understanding and concurrent processing framework based on the message bus and the container virtualization technology is proposed.The distributed message bus is used to cache and distribute video processing data/tasks,ensuring the smooth and reliable operation of the video processing.The concurrent processing of video content understanding function is realized by two mechanisms of task parallel and model parallel.Thus,the efficiency of video processing is improved,and it can support seamless access and large-scale deployment of video content artificial intelligence(AI)processing function module.

关 键 词：视频内容理解 消息总线 容器 虚拟化 并发处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]