preview

The Problem Of Analysing Behaviours In Video Has Been The

Decent Essays

The problem of analysing behaviours in video has been the focus of several researchers’ efforts and several systems have been described in the literature. Action recognition is the process of labelling image sequences with action labels. The task is challenging due to variations in motion performance, recording settings and inter-object differences. Generally, the action recognition process can be divided into two steps 1st feature extraction and representation and 2nd action class prediction. The first step deals with the extraction and encoding of features to describe motions of interest. Multiple features might be extracted for motion modelling prediction. Many techniques have been used in this step including parametric models, …show more content…

Local feature methods are entirely based on descriptors of local regions in a video, no prior knowledge about human positioning nor of any of its limbs is given. In the following subsections, these categories are discussed in further. Fig 7: Action recognition methods

6.1 Human Model Based Methods
Human model based methods recognize actions by employing information such as body part positions and movements. A significant amount of research is dedicated to action recognition using trajectories of joint positions, body parts, or landmark points on the human body with or without a prior model of human kinematics. Approaches in this field depend on a previous psychophysical work on visual interpretation of biological motion. This work shows that humans are able to recognize actions from the motion of a few moving light displays attached to the human body. The localization of body parts in movies has been investigated in the past and some works have shown impressive results. However in general, the detection of body parts is a difficult problem in itself, and results are still limited especially for the case of realistic and less constrained video. Some recent approaches try to improve their results by assuming particular motion patterns, hence improving body parts tracking. However, this also limits their application to action

Get Access