Overview
Real-Time video Analytics Engine is a Platform that utilizes advanced image processing algorithms to turn video into actionable intelligence. Large Scale Video Processing is the greatest challenge holding a great potential for Analytics. Real-Time Video Analytics can be used for Monitoring Traffic Control, Retail Store Monitoring, Surveillance and Security.
Don's Real-Time Video Analytics Platform is real-time, low cost, accurate analysis of live videos using Open Source Big Data Technologies including OpenCV, Apache Kafka, Apache Spark, HDFS, Tensor Flow, and Amazon S3.
Business Challenge
-
Traditional Computer Vision System suffers from a limitation, that a server along with CV library collects and process the data at the same time. A failure in the server causes loss of Streaming Video Data.
-
Detecting a node failure and switching the processing to another node may result in fragmentation of data.
-
Major challenge relies on running an image processing algorithms in a dynamic environment.
Thus managing and efficiently analysing this data brings us with a challenge to build a system which can eliminate the addressed problems.
Solution Offered
Don Team came up with a solution to perform real-time video analytics backed by open source big data technologies.
The system is divided into three main components -
-
Video stream collector
-
Stream data buffer
-
Video stream processor