A Proposed Technique for Automatic Recognition of Human Activities

Abstract

While computer vision is widely employed in a wide variety of applications, the precise and efficient identification of human behaviour remains a challenging area in computer vision science. Recent research has concentrated on smaller issues such as approaches for human action recognition of depth data, 3D skeleton data, photographic data, spatiotemporal methods focusing on interest and the identification of human activity. Despite this, no systematic survey of human behaviour appraisal has been conducted. To that end, we present a comprehensive review of methods for identifying human actions, including advances in the hand design of action characteristics in RGB and depth data, established methods for representing deeper learning action-based features, advancements in the methodology for identifying human-object interaction, and prominent present methods of deeper information.