Computer vision has been attracting interest throughout the years and is an active research topic. One of the major interests in computer vision is to attempt to detect and recognise actions or activities of the objects from a video sequence. Significant importance has been attached since the number of cameras has been increasing throughout the years.

Lara Schembri has just completed her research in this interesting field of ICT.

Action recognition has several applications, such as highlighting and storing important frames, gaming, security and surveillance and archiving. One example of this is the Microsoft Kinect, which involves a camera recognising a human and mimicking any of the movement on the game. Another example is to use gesture recognition to interface with the computer without the means of a touch screen.

Ms Schembri’s research presents the different algorithms which could be used for the action recognition process.

The process consists of recognising basic action such as walking, running, jogging, clapping, waving and boxing, where the structure of the action recognition process consists of background subtraction, feature extraction and recognition. The recognition techniques used were the Hidden Markov Model (HMM), which is a time variant system, and the Support Vector Machine (SVM), a binary classifier.

This research was possible thanks to a STEPS scholarship which is part-financed by the EU’s European Social Fund.