Tensorial modeling of dynamical systems for gait and activity recognition

点击翻译按钮获取中文摘要

Biometrics such as face, iris, or fingerprint recognition have received growing attention in the last decade, as automatic identification systems for surveillance and security have started to enjoy widespread diffusion. They suffer, however, from two major limitations: they cannot be used at a distance, and require user cooperation, assumptions impractical in real-world scenarios. Interestingly, psychological studies show that people are capable of recognizing their friends just from the way they walk, even when their gait is poorly represented by point light display. Gait has several advantages over other biometrics, as it can be measured at a distance, is difficult to disguise or occlude, can be identified even in low-resolution images, and is non-cooperative in nature. Furthermore, gait and face biometrics can be easily integrated for human identity recognition.Despite its attractive features, though, gait identification is still far from being ready to be deployed in practice. What limits its adoption in real-world scenarios is the influence of a large number of nuisance factors which affect appearance and dynamics of the gait. These include, for instance: walking surface, lighting, camera setup (viewpoint), but also footwear and clothing, objects carried, time of execution, walking speed. Similar issues are shared by other applications of motion classification, such as action and activity recognition. Multilinear or tensorial models, in which a number of (nuisance) factors linearly mix to generate what we observe (in our case the walking gait), have been proven in the recent past to be able to describe the influence of such factors, for instance in the context of face recognition. However, video sequences are more complex objects than single images. We first need to represent video footages in a compact way.Encoding the dynamics of videos by means of some sort of dynamical model has been proven effective in both action recognition and gait identification, in situations in which the dynamics is critically discriminative. Besides, the actions of interest have to be temporally segmented from a video sequence, while actions of sometimes very different lengths might have to be compared. Dynamical representations are very effective in coping with temporal detection and compression, and indeed several researchers have explored the idea of encoding motions via linear, nonlinear, stochastic or chaotic dynamical systems.In this project, therefore, we propose to develop a novel, general framework for the classification of video sequences (with a focus on the walking gait), based on the application of tensorial decomposition techniques to image sequences represented as realizations of suitable dynamical models.The proposed framework will allow us to deal with the issue of the nuisance factors which greatly affect identification from gait and activity recognition in a principled way. The main goal is to push towards a more widespread diffusion of gait ID, as a concrete contribution to enhancing the security levels in the country in the current, uncertain scenarios. With their implications for crime prevention and security, biometrics and surveillance are fast growing business areas, a fact reflected by the increasing number of government-sponsored initiatives in the area in most advanced economies. In addition, the techniques devised in this proposal are extendable to action and identity recognition with immense commercial exploitation potential, ranging from content-based video retrieval from repositories such as YouTube, to HMI, to interactive video games, etcetera.