Detection and Identification of Behaviors and Biometric Data

To Develop An Integrated System for Detection and Identification of Behaviors and Bio-metric Data

Due to advances in Information Technologies, currently it is easy to collect, store and analyze data. In particular, a recent interest has been given to the use of 3D video cameras, which are becoming more affordable and more universally used. This practical research is performed within this context and assumes an interdisciplinary work, between the several Computer Science areas, such as Computer Vision, Human-Computer Interaction and Artificial Intelligence (with a particular emphasis on Data Mining). A synopsis of the related background is described below.

One interesting research and recent trend relies in identification of human activity from video (Turaga, 2008). While several temporal and special models have been proposed to identify these activities (Weinland el al., 2011), there are very few relevant studies that use 3D video as a biometric technique. These few studies focus more on using only one biometric aspect, such as use of facial expressions (Bowyer et al., 2006) or walking patterns (Yam et al., 2004), thus lacking the ability to integrate multiple human signals.

This practical research aims at developing a real-time system capable of analyzing 3D video, under a multi-model biometric approach (Jain, 2004). The goal is to develop a prototype system that can support the management of statistical information related with human behavior, which is vital for decision-making. For instance, such system could be set inside a commercial store window, to assess if a given human is a potential costumer, neutral person or a possible threat (e.g. thief). A particular focus is given towards face and gesture recognition. Several studies have addressed this issues separately, such as: face detection - (Lee et al., 2003)(Zhao et al., 2003); and gesture recognition – (Liu et al., 2009)(Fenn, 2010). It should be noted that video gesture recognition is a very recent trend, since previous works, such as (Liu et al., 2009), are mostly based on sensors (e.g. accelerometers). Moreover, in the novelty of this research is that a multi-modal approach is assumed, where several biometric aspects, such as gender, face and gesture aspects will be integrated, in order to achieve novel capabilities in terms of interesting human behavior detection.

This is performed within this context and assumes an interdisciplinary work, between the several Computer Science areas, such as Computer Vision, Human-Computer Interaction and Artificial Intelligence (with a particular emphasis on Data Mining).

Data protection in data processing procedures is best adhered to when it is already integrated in the technology when created.

There are two major research methodologies that can guide this research. Given the novelty and nature of the problem to be addressed, an important part of the research will adopt an exploratory research approach (Jaeger and Halliday, 1998), where several Computer Vision and Artificial Intelligence techniques will be tested to check what sort of useful human behaviors can be detected within a given scenario (e.g. commercial store). On the other hand, since this research assumes de development of a prototype, which can be seen as a the proposal of a novel artifact, this work will follow a Design Science research methodology (Vaishnavi and Kuechler, 2004). Regarding the
application of Data Mining techniques, these will be guided by the CRISP-DM methodology (Chapman et al., 2000).

This practical research is structured into four major tasks: 1) gender detection; 2) facial recognition and detection of human behaviors; 3) Implementation of the system prototype; and 4) testing the implemented
prototype in a real-environment which is quite user-friendly.

In this practical research a multi-modal approach is assumed, where several biometric aspects, such as gender, face and gesture aspects will be integrated, in order to achieve novel capabilities in terms of interesting human behavior detection.

The goal is to develop a prototype system that can support the management of statistical information related with human behavior, which is vital for decision-making. For instance, such system could be set inside a commercial store window, to assess if a given human is a potential costumer, neutral person or a possible threat (e.g. thief).

Once this practical research is into practice or use, it can be scaled up through out the country for better identification and protection.

Jhpiego & FHI360 - ICT4D Specialist.

Vera Edu. Consultancy - President.

Professional Programming and ICT4D Skills.

NA

I'm looking for funds to put this into practical.

Limited Available Budgets and a grant/funding can overcome this.