人工智能视频深度处理技术的电力施工监测

Power construction monitoring based on artificial intelligence video deep processing technology

针对云计算框架处理电力施工监测视频数据延迟高、功耗大等问题,该研究通过设计mVideo系统,借助于移动边缘节点捕获电力施工监测视频,在边缘节点部署分区深度神经网络计算模型以深度处理视频流,传输模块可以将每个节点的中间结果和状态信息发送到存储服务器中。最终得到mVideo系统比云计算框架处理视频数据的执行时间缩短了45%,功耗降低了113%,证明了mVideo系统在上述部分指标方面的优势,更加实用可靠。

Regarding the high latency and high power consumption of the power construction monitoring video data of the cloud computing framework,this research uses the design of the mVideo system to capture power construction monitoring videos with the help of mobile edge nodes,and deploys partitioned deep neural network computing models at the edge nodes for in-depth processing,meanwhile,the transmission module can send the intermediate results and status information of each node to the storage server.Finally,compared with the cloud computing framework,the execution time of the mVideo system to process video data is shortened by 45%,and the power consumption is reduced by 113%,which proves that the mVideo system has the advantages of some of the above indicators and is more practical and reliable.