Article: Development of detection and recognition of human activity in sports using GMM and CNN algorithms Journal: International Journal of System of Systems Engineering (IJSSE) 2024 Vol.14 No.5 pp.520 - 538 Abstract: The system offers a comprehensive mechanism for tracking several individuals and measuring their combined actions. Our method assumes that a person's mobility, activity, and neighbouring people's motions and behaviours are meaningfully interconnected. We propose a hierarchy of activity types to allow a more natural transition from solo mobility to communal motion. The approach provides a two-tiered hierarchical graphical model to learn the spatial and temporal links between tracks, tracks, and activity segments. We also propose combining conviction engendering with a branch and bound methodology modified with whole number programming to solve this intractable joint inference problem. This work uses motion and context data to jointly model and detect associated movie actions. The realisation that geographically and temporally related events rarely happen separately and often serve as backdrops prompted this. A hierarchical two-layer conditional random field model represents action segments and activities. The model integrates motion and backdrop variables at numerous levels and generates statistics that automatically identify typical patterns. Inderscience Publishers - linking academia, business and industry through research

Title: Development of detection and recognition of human activity in sports using GMM and CNN algorithms

Authors: S. Dhivya Karunya; Krishna Kumar

Addresses: Department of Electronics and Communication Engineering, SEA College of Engineering and Technology, Bengaluru, Karnataka, India ' Department of Electronics and Communication Engineering, Gopalan College of Engineering and Management, Bengaluru, Karnataka, India

Abstract: The system offers a comprehensive mechanism for tracking several individuals and measuring their combined actions. Our method assumes that a person's mobility, activity, and neighbouring people's motions and behaviours are meaningfully interconnected. We propose a hierarchy of activity types to allow a more natural transition from solo mobility to communal motion. The approach provides a two-tiered hierarchical graphical model to learn the spatial and temporal links between tracks, tracks, and activity segments. We also propose combining conviction engendering with a branch and bound methodology modified with whole number programming to solve this intractable joint inference problem. This work uses motion and context data to jointly model and detect associated movie actions. The realisation that geographically and temporally related events rarely happen separately and often serve as backdrops prompted this. A hierarchical two-layer conditional random field model represents action segments and activities. The model integrates motion and backdrop variables at numerous levels and generates statistics that automatically identify typical patterns.

Keywords: CNN; convolutional neural network; GMM; Gaussian mixture model; sports activity; medicine; athletics; entertainment business; machine learning; human activity in sports.

DOI: 10.1504/IJSSE.2024.140756

International Journal of System of Systems Engineering, 2024 Vol.14 No.5, pp.520 - 538

Received: 12 Mar 2023
Accepted: 17 Apr 2023
Published online: 02 Sep 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Development of detection and recognition of human activity in sports using GMM and CNN algorithms

Keep up-to-date