Presenter: Dr. Vassilis Athitsos

Affiliation: The University of Texas at Arlington, Department of Computer Science and Engineering

Title: Human Motion Analysis Using Computer Vision: Some Applications,
Methods, and Challenges

Abstract: There are numerous real-world applications that can benefit from computer vision methods performing human motion analysis. This talk will discuss some of the human motion analysis projects that we have been pursuing over the last several years, including sign language recognition, and automatic scoring of physical exercises. These projects have presented several important challenges, that are commonly encountered in human motion analysis. One such challenge is the lack of adequate amounts of training data, that necessitates the use of machine learning methods that can be effective with only a few examples of training objects per class. Another challenge is the difficulty of accurate detection and tracking of individual body parts, which necessitates the development of recognition modules that are robust to detection and tracking errors. A third challenge is matching the typical human motion analysis paradigms developed by the computer vision community with the needs of actual applications, as oftentimes the assumptions made in the lab are not suitable for applications in the real world. The talk will discuss methods that we have developed, results on our target applications, and remaining open problems.

Bio: Vassilis Athitsos received a PhD degree in computer science from Boston University in 2006. Since August 2007 he is a faculty member at the Computer Science and Engineering department at the University of Texas at Arlington, where he currently serves as a full professor. His research interests include computer vision, machine learning, and data mining. His recent work has focused on gesture and sign language recognition, detection and tracking of humans using computer vision, and automated scoring of physical exercises performed by children. His research has been supported by numerous grants from the National Science Foundation, including an NSF CAREER award.

Presenter: Dr. Reuth Mirsky

Affiliation: The University of Texas at Austin, Computer Science Department.

Abstract:
Existing pervasive computing applications entail high degrees of communication, but existing deployments by and large rely on either a backend Internet connection to support communication or provide only one-way distribution of data (e.g., via environmental beacons). However, the pervasiveness of sensing, computation, and communication has changed the landscape of potential pervasive crowd applications. The plain “chattiness” of our everyday environments opens broad new possibilities for pervasive computing devices to opportunistically leverage each other, where the potential capability of a whole opportunistic crowd is much larger than the sum of its individuals’ capabilities. In this talk, I will present both concrete motivating application domains and technical capabilities and constraints that lead us to consider the potential of direct device-to-device collaboration in support of crowd applications in pervasive computing. Drawing on my group’s work in collaborative, opportunistic context-awareness, I will create a roadmap for research in support of a future vision of opportunistic crowds

Presenter: Dr. Christine Julien
Dr. Julien is a professor in the Center for Advanced Research in Software Engineering (ARiSE) in the Department of Electrical and Computer Engineering at the University of Texas at Austin, which she joined in 2004. She is the director of the Mobile and Pervasive Computing Group, where her research focuses on the intersection of software engineering and dynamic, unpredictable networked environments. Her specific focus is on the development of models, abstractions, tools, and middleware whose goals are to ease the software engineering burden associated with building applications for pervasive and mobile computing environments. Dr. Julien’s research has been supported by the National Science Foundation (NSF), the Air Force Office of Scientific Research (AFOSR), the Department of Defense, and Freescale Semiconductors. The work has been recognized by an NSF CAREER award and an AFOSR Young Investigator Award, and the results have appeared in many peer reviewed journal and conference papers. Dr. Julien graduated with her D.Sc. in 2004 from Washington University in Saint Louis, where her doctoral research under the supervision of Dr. Gruia-Catalin Roman focused on developing a middleware called EgoSpaces that provided an intuitive data-structure abstraction to support application coordination in mobile computing environments. She earned her M.S. degree in 2003 and her B.S. with majors in Computer Science and Biology in 2000 (both also from Wash. U.).

This Research Experience for Undergraduates (REU) Site engages students in research in the emerging area of Smart & Connected Communities (S&CC), the essential building blocks of smart cities.

Through a nationwide recruitment process, cohorts of 10 undergraduate students are selected to participate in a ten-week summer research program, for three summers, hosted at Texas State University. Special efforts are made to recruit students from primarily teaching universities and colleges while ensuring the participation of women, veterans, and minorities. The project includes mentoring by experienced computer science faculty members, technical seminars and workshops, student presentations, and professional development opportunities. These activities are carried out in an intellectual environment that develops the students’ professional careers by enhancing their technical, communication, and social skills.

Research

The research component in this REU Site falls along three main thrusts in building the next generation of information and communication technologies supporting the development of S&CC:

1. Developing models, mechanisms, and tools for smart human-environment interactions to monitor and protect smart infrastructure against accidental failures and malicious attacks.
2. Developing new methods for collecting and analyzing various types of data ranging from human biosignals to activity and crowd-sourcing data, to promote the health and well-being of citizens in urban settings.
3. Developing novel scheduling and monitoring methods for intelligent transportation systems that would decrease congestion, promote safety and reduce adverse environmental impacts in urban settings. Evaluation of the proposed methods is done through theoretical and numerical analysis, simulation experiments, and real implementations with various devices, sensors, and robots.