A Survey of pet Action Recognition with action Recommendation based on HAR

Abstract

Human Action Recognition (HAR) is a challenging task, the aim of HAR is to detect the person performing the action in an unknown video sequence, to determine the duration of the action and to identify the type of action. In sports, the main idea of HAR is to monitor the performance of players, i.e. to detect players, track their movements, identify the actions performed, compare various actions, compare the performance of different types and skills, or perform automatic statistical analysis, an overview of HAR research focuses on various approaches performed on publicly available datasets, including actions for everyday activities. This paper raises the idea of whether pet action recognition can be linked to HAR. When pet action can be recognised, the corresponding data can be used to target recommendations for the health and diet of the pet. pet and human movement recognition are closely related and this paper highlights and summarises HAR as a contribution to future research.