Recognition of Human Activity Using Deep Learning Models

Read full text Summarise with VIRAT
Authors: A. Ranjeeth, P. Yogapriya, S. Keerthi, N. Sengeniammal
Cited by: 0
References: 12

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Human Activity Recognition aims to automatically identify human actions using video, wearable sensors, or environmental data.It is widely used in hospitals, schools, public areas, malls, and smart environments for safety monitoring, anomaly detection, behavior understanding, and automation.Traditional HAR systems mainly relied on handcrafted features, which made them sensitive to real-world challenges such as lighting changes, occlusions, background variations, and complex human movements.Because of this, their performance often dropped in uncontrolled settings.With the rise of deep learning, HAR has become more accurate and robust.Modern models such as CNNs, RNNs, 3D-CNNs, GCNs, and transformers can automatically extract spatial temporal features, understand motion patterns, and handle multi-actor scenarios more effectively.This paper is a survey that reviews recent deep learning-based HAR methods.Instead of introducing a new model, we provide an organized overview of existing techniques, discuss their strengths and weaknesses, and highlight important research trends and open challenges.The goal is to guide future work and support the development of more efficient HAR systems.

Fields of Study & Subjects

Human Pose and Action Recognition Context-Aware Activity Recognition Systems Hand Gesture Recognition Systems

Publication details

Journal / source
Advances in intelligent systems research/Advances in Intelligent Systems Research
Pages
145–151
Publication date
1 April 2026
Language
EN
Document type
Book chapter