Skip to main content

Automatic Recognition of Human Activities in Surveillance Videos: Overcoming the Curse of Dimensionality

Funding: 2009: $95,000
2010: $90,000
2011: $50,000

Project Member(s): Piccardi, M.

Funding or Partner Organisation: Australian Research Council (ARC Discovery Projects)

Start year: 2009

Summary: Automated recognition of certain human activities in surveillance videos can prove critical for the security and safety of environments. To date, the effectiveness of automated activity recognition has been limited by ''''the curse of dimensionality'''' that is the difficulty of learning adequate activity models in high-dimensional spaces. This project will propose novel non-parametric approaches and dimensionality-reduction techniques to tackle this problem simultaneously from two sides. The outcome of this project will be an activity recognition technology suitable for use in real environments.

Publications:

Bargi, A, Xu, YD & Piccardi, M 2018, 'AdOn HDP-HMM: An Adaptive Online Model for Segmentation and Classification of Sequential Data', IEEE Transactions on Neural Networks and Learning Systems, pp. 3953-3968.
View/Download from: UTS OPUS or Publisher's site

Zhang, G & Piccardi, M 2015, 'Structural SVM with Partial Ranking for Activity Segmentation and Classification', IEEE SIGNAL PROCESSING LETTERS, vol. 22, no. 12, pp. 2344-2348.
View/Download from: UTS OPUS or Publisher's site

Keywords: Computer Vision, Video Surveillance, Nonparametric techniques, Dimensionality reduction,

FOR Codes: Commercial security services, Pattern Recognition, Computer Vision, Pattern Recognition and Data Mining, Property Services (incl. Security), Computer software and services not elsewhere classified, Computer Time Leasing, Sharing and Renting Services