Improving the efficiency of streaming video processing through the use of serverless technologies

Abstract

This article analyzes the approaches to solving the problem of analyzing the input video stream in real-time. The architectures for building a video stream processing system in which more than one video stream source operates, transmitting data simultaneously in real-time to solve the problem of frame-by-frame recognition, and propose a solution based on the conditions of dynamic change in the number of sources, real-time processing of a separate source. Implementation with statically allocated resources is considered, and also serverless architecture solving the revealed shortcomings is offered. The speed and modularity of the proposed architecture solve the identified shortcomings of the architecture with statically allocated resources and also allow to reduce downtime of computing power by automatically quickly dynamically removing resources that are not used at a particular time or adding additional resources with increasing load. Ref. 3, pic. 3.

Keywords

• video analytics
• static computing resources
• serverless functions
• frame-byframe video analysis