Dynamic Video-Based Activity Recognition using 3D Convolutional Neural Network

7 Pages Posted: 13 May 2022

See all articles by Advait Naik

Advait Naik

Vivekanand Education Society’s Institute of Technology Mumbai, India

Roshnee Matlani

Vivekanand Education Society’s Institute of Technology Mumbai, India

Date Written: April 8, 2022

Abstract

Human activity recognition has become a topic of conversation in automated video surveillance and computer vision research in recent years. Surveillance systems with activity recognition algorithms are necessary to prevent treacherous activities from arising in particular. The research work introduces 3D Convolutional Neural Networks (3D CNN) with a spatial-temporal dynamic movement for activity recognition. The research provides a unique 3D CNN methodology for activity recognition by extracting kinematic motion information stored in consecutive neighboring frames and utilizing retrieves parameters from both the space and time dimensions. The experimental evaluation is performed using the benchmark KTH and Weizmann datasets, which describe the activities of persons in daily lives and work. There are numerous existing methods for activity recognition, therefore the proposed approach is compared to previously proposed methods and it is found that the proposed methodology vanquishes previously reported results with an accuracy of 97.52 % on the KTH dataset and 98.6 % on the Weizmann dataset.

Keywords: Human activity recognition, 3D Convolutional Neural Networks (3D CNN), Video surveillance, KTH dataset, Weizmann dataset

Suggested Citation

Naik, Advait and Matlani, Roshnee, Dynamic Video-Based Activity Recognition using 3D Convolutional Neural Network (April 8, 2022). Proceedings of the 7th International Conference on Innovations and Research in Technology and Engineering (ICIRTE-2022), organized by VPPCOE & VA, Mumbai-22, INDIA, Available at SSRN: https://ssrn.com/abstract=4108889 or http://dx.doi.org/10.2139/ssrn.4108889

Advait Naik (Contact Author)

Vivekanand Education Society’s Institute of Technology Mumbai, India ( email )

Roshnee Matlani

Vivekanand Education Society’s Institute of Technology Mumbai, India ( email )

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
58
Abstract Views
296
Rank
671,302
PlumX Metrics