#object classification

Video analytics framework detection performance is worked at cloud. Object detection and classification are the basic tasks in video analytics and become the initial point for other complex submissions. Old-fashioned video analytics approaches are manual and time consuming. These are particular due to the very participation of human factor. This paper present a cloud based video analytics framework for accessible and robust analysis of video streams. The framework enables an operative by programing the object detection and classification process from recorded video streams. An operative only specifies an analysis criteria and period of video streams to analyze. The streams are then realized from cloud storage, cracked and analyzed on the cloud. The framework performs compute severe parts of the analysis to CPU powered servers in the cloud. Vehicle and face finding are accessible as two case studies for assessing the framework, with one month of data and a 15 node cloud. The framework consistently performed object detection and classification on the data, comprising of 21,600 video streams and 175 GB in size, in 6.52 hours. The GPU enabled placement of the framework took 3 hours to perform analysis on the same number of video streams, thus making it at least double as fast than the cloud deployment Without GPUs. The analysis framework is high.