Pattern classification of human interactions from videos / Muhsin Abdul Mohammed

Muhsin , Abdul Mohammed (2018) Pattern classification of human interactions from videos / Muhsin Abdul Mohammed. Masters thesis, University of Malaya.

	PDF (The Candidate's Agreement) Restricted to Repository staff only Download (818Kb)
Preview	PDF (Thesis M.A) Download (1753Kb) | Preview

Abstract

The objective of this research project is to build a machine learning model to classify human interactions from a stream of video. Being able to classify human interaction from videos is essential in the development of robotic assistance systems, video annotation, surveillance systems and many more applications. It is necessary that the algorithm performing this task needs to be robust and only relies on monocular vision systems. In order to build a classifier capable of achieving this task, the machine learning model needs to be able to learn spatial and temporal patterns from the videos. A cascaded architecture of Convolutional Neural Networks and Recurrent Neural Networks have been created to achieve this task in this research. There have been investigations made to identify the best spatial and temporal architectures that would give the optimal result.

Actions (For repository staff only : Login required)