Arslan Basharat, Ph.D.

 

 

Current Affiliation: Kitware, Inc.

 

Advisor: Dr. Mubarak Shah

 

Computer Vision Laboratory

Room: HEC-254

School of Electrical Engineering and Computer Science

University of Central Florida

Orlando, FL

 

e-mail:

 

 


[Short Bio] [Research Projects] [Publications] [Misc. Links]


Short Bio

    Work Experience

    Education

    Research Interests

Research Projects

[^ Back to Top]

Learning Scene Model from Surveillance Videos
[Project Page Dataset Shared] [Related Publication]  
We investigate new methods to integrate a higher-level layer to the traditional surveillance pipeline to perform anomalous event detection and scene model feedback. A novel framework for learning patterns of motion and size of objects in static surveillance cameras is presented. Each pixel is modeled as a multivariate Gaussian Mixture Model (GMM) of motion (destination location & transition time) and size (width & height) parameters of the objects at that location. We also show the use of this scene model to improve object detection through pixel-level parameter feedback of the minimum object size and background learning rate.
Content Based Video Matching Using Spatiotemporal Volumes
[Project Page Dataset Shared] [Related Publication]  
This project focused on matching telecast videos based on a new method of object level matching. We extract foreground and background spatiotemporal volumes from the video through motion segmentation. This approach relies on SIFT interest point trajectories and homography based motion segmentation. Multiple features (color, texture, motion, & interest point descriptors) are used to represent and match volumes. Earth Mover's Distance (EMD) is used as the metric for feature comparison, while a graphical model is used for volume correspondence.  Experiments have been performed on videos from BBC Motion Gallery, TRECVID, and Google Video.
Chaotic Invariants for Human Action Recognition
[Project Page] [Related Publication]                           

[^ Back to Top]

This project explores the feasibility of theory of chaotic systems for solving different vision problems. As a first step, we have used it to model and analyze nonlinear dynamics of human actions, where trajectories of body joints are used as the representation of the nonlinear dynamical system generating the action. Using these trajectories, each action is represented in terms of Chaotic Invariants which measure the dynamical and metric structure of the reconstructed phase spaces corresponding to each trajectory.
ICCV 2005 Computer Vision Contest: Where Am I?              (Honorable mention)
[Project Page]  
This contest was held at the International Conference of Computer Vision, 2005. Given a set of images taken at known GPS locations, the problem was to find GPS locations of test images taken at unknown locations but roughly in the same area (view overlap). Our team consisted of five researchers including myself. We received an honorable mention at the conference for achieving fourth place amongst twenty participating teams from top universities all over the world.
Automated Visual Inspection of Railroads
[Project Page]                                                        

[^ Back to Top]

The purpose of this research project is to develop an automated visual inspection system for reliable identification and localization of structural defects in railroad tracks. The system relies on computer vision and pattern recognition techniques to detect abnormal parts along the track. This is useful for the relevant personnel to guarantee railroad compliance with the Federal Railroad Administration Track Safety Standards. This project is funded by Florida Department of Transportation.
 
Visual Monitoring of Railroad Grade Crossing
[Project Page]  
This included design and testing of a surveillance system with self sustained solar power and wireless communication. It is used for raising live alarms remotely on the project website and locally to warn the individuals. The events of interests in this project were: incoming train detection, person entering danger zone, person staying on track, etc. I successfully led a team of three members to take this surveillance project through the final stages of development, testing, and delivery.
COCOA: Alignment, Object Detection, Object Tracking and Indexing of Aerial Videos
[Project Page]

[^ Back to Top]

I helped in this project by implementing a blob tracker module for COCOA in Java. The tracker worked on the output of frame registration and object detection modules. It used the spatial and appearance model similar to that in KNIGHT tracking engine. Details of the project are available on the project page.

 

Publications

[^ Back to Top]

 

Arslan Basharat and Mubarak Shah,

IEEE Intl. Conf. on Computer Vision (ICCV), Oct 2009

[PDF][Project Page]

 

Alexei Gritai, Arslan Basharat, and Mubarak Shah,

International Conference on Pattern Recognition (ICPR), Dec 2008

(Oral presentation: ~18% acceptance rate)

[PDF][Project Page]

 

Arslan Basharat, Alexei Gritai, and Mubarak Shah,

IEEE Conf. of Computer Vision and Pattern Recognition (CVPR), June 2008

[PDF][Project Page]

 

Arslan Basharat, Yun Zhai, and Mubarak Shah,

Journal of Computer Vision and Image Understanding (CVIU), Volume 110, Issue 3, June 2008, Pages 360-377

[PDF] [Presentation (pdf)] [Project page]

 

Saad Ali, Arslan Basharat, and Mubarak Shah,

IEEE International Conference on Computer Vision (ICCV) 2007

[PDF] [Presentation (pdf)] [Project page]

 

Andrew Miller, Arslan Basharat, Brandyn White, Jingen Liu, and Mubarak Shah,

CLEAR Evaluation Campaign and Workshop, Baltimore MD, 2007

[PDF]

 

Mubarak Shah, Asaad Hakeem, and Arslan Basharat,

Publication of SPIE, March 2006

[PDF]

 

Jingen Liu, Yun Zhai, Arslan Basharat, Bilal Orhan, Saad M. Khan, Humera Noor, Phillip Berkowitz, and Mubarak Shah,

TREC Video Retrieval Evaluation Forum (TRECVID) 2006 

[PDF]

 

Arslan Basharat, Asaad Hakeem, Mubarak Shah, and Abhijit Mahalanobis,

OE Magazine, Member Publication of SPIE, November 2005

[PDF]

 

Yun Zhai, Jingen Liu, Xiaochun Cao, Arslan Basharat, Asaad Hakeem, Saad Ali, Mubarak Shah, Costantino Grana and Rita Cucchiarra,

TREC Video Retrieval Evaluation Forum (TRECVID) 2005

[PDF]

 

Arslan Basharat, Necati Catbas, and Mubarak Shah,

Workshop on Sensor Networks and Systems for Pervasive Computing (PerSeNS)

o        In conjunction with International Conference on Pervasive Computing and Communications (PerCom) 2005

[PDF][Presentation (pdf)]

 

Necati Catbas, Mubarak Shah, J. Burkett, and Arslan Basharat,

The 4th International Workshop on Structural Control (4IWSC), 2004

[Proceedings]

 

Misc Links

[^ Back to Top]

[^ Back to Top]