Video-based Human Action Recognition using Deep Learning: A Review

Abstract

Human action recognition is an important application domain in computer vision. Its primary aim is to accurately describe human actions and their interactions from a previously unseen data sequence acquired by sensors. The ability to recognize, understand and predict complex human actions enables the construction of many important applications such as intelligent surveillance systems, human-computer interfaces, health care, security and military applications. In recent years, deep learning has been given particular attention by the computer vision community. This paper presents an overview of the current state-of-the-art in action recognition using video analysis with deep learning techniques. We present the most important deep learning models for recognizing human actions, analyze them to provide the current progress of deep learning algorithms applied to solve human action recognition problems in realistic videos highlighting their advantages and disadvantages. Based on the quantitative analysis using recognition accuracies reported in the literature, our study identifies state-of-the-art deep architectures in action recognition and then provides current trends and open problems for future works in this filed.

Delving Deep Into One-Shot Skeleton-Based Action Recognition With Diverse Occlusions

“…action recognition [8] stem from multiple factors: complex backgrounds, variations in human body shapes, changing viewpoints, or motion speed alterations. In contrast to videobased action recognition [9]- [11], skeleton-based action recognition is less sensitive to appearance factors and has the advantage of superior efficiency by using sparse 3D skeleton data as input, which ensures fast inference speed and small memory usage. Thanks to the advancement of depth sensors [12] and lightweight and robust pose estimation algorithms [13], [14], obtaining high-quality skeleton data is becoming easier.…”

Design and Development of an Imitation Detection System for Human Action Recognition Using Deep Learning

“…Shu et al in [ 6 ] presented the Omni-Training framework, which connects pretraining and metatraining for effective few-shot learning with limited data. These algorithms can learn complex patterns and relationships among different features of human actions, making them more effective in recognizing and classifying actions than traditional ML approaches [ 7 ].…”

“…Then, these features are used to recognize actions. As a result, recognizing an action based on features can be viewed as a classification problem [ 7 ].…”

Detecting Human Actions in Drone Images Using YoloV5 and Stochastic Gradient Boosting

“…However, extending this task to drone-captured images and videos is an emerging topic. Human action recognition is a well-studied problem which is categorized into (i) posebased [2][3][4], (ii) single-image-based [5][6][7], and (iii) video-based action recognition [8][9][10]. However, detecting actions in single images is a less explored area because it faces the problem of the unavailability of annotated temporal data for action detection [11][12][13].…”

“…Pareek et al [9] briefly covered the human action recognition techniques using machine learning and deep learning methods for the years 2011-2019. Pham et al [10] presented most important deep learning models for action recognition, analyzed their performance, and then discussed future prospects and challenges for recognizing actions in realistic videos. Rohrbach et al [11] prepared a dataset for kitchen activities.…”