Deep learning approach to Automated data collection and processing of video surveillance in sports activity prediction

Abstract

Human activity recognition is one of today's key fields of automated video surveillance. The technology of smart surveillance technology plays a crucial role. Despite efforts in recent years, it is still difficult to recognize human behaviors from live video. Human activity can vary from basic behaviors to complicated behaviors. Depth cameras currently released have an efficient 3D estimate of body connecting locations in the temporal depth map collection. This article proposed a method for recognizing human behavior and considered the challenge of achieving a descriptive marking of activities by labeling individual sub-activities. The behaviors take place over a long period and have many sequential sub-activities. A sports activity prediction of video surveillance framework is proposed in this article. The suggested operation descriptor considers the sequence classification challenge to be the behavior recognition problem. Deep Learning is used to detect human behaviors in the proposed method. The method is tested on two regular identification benchmark functions. Effects of the research revealed that the solution developed exceeds cutting-edge methodologies.