HomeProjectsPublicationsPeopleLinks

  Home > Projects  
         
 


 


We have developed a crawler for Google Earth which can segment nucleated villages in satellite imagery. [more] 

Partially funded by LUMS Computer Science Department

  
         
 


 


We are exploring structure from motion for the case where the scene points deform nonrigidly. Assuming an orthographic camera, the key to our approach is to model the 3D structure as a linear combination of basis trajectories. [more]

Partially funded by Higher Education Commission, Pakistan and LUMS Computer Science Department

  
     
 
 
 

Correlation coefficient is a robust similarity measure but is expensive to compute. We have developed three different elemination strategies to speed up the peak-search process for correlation-coefficient based template matching. [more]

Partially funded by LUMS Computer Science Department
 
     
 
 
 

A collaborative project with Dr Ashfaq Khokhar and Dr Dan Schonfeld at the Univ of Illinois at Chicago, the goal is to develop large scale trajectory datasets and to evaluate the performance of current trajectory classification algorithms as the number of classes scale up. [more]

Partially funded by National Science Foundation, USA and LUMS Computer Science Department
 
     
 
 
 

We are developing a low-cost ultrasound training simulator which will eliminate the need for trainee doctors to practice on real patients. [more]

Funded by National ICT R&D Fund, Pakistan
  
     
 
 
 


We are developing robust tracking and analysis techniques for floresence and phase-contrast microscopic assays, with special emphasis on occlusion resolution. This work is in collaboration with Dr Shahid Khan (LUMS SSE) and Dr Justin Molloy (NIMR, London)

Funded by LUMS Computer Science Department and British Council Link Program

  
     
 
 
 


In video codecs, block based motion compensation is used to reduce temporal redundancy. We show that if the difference between the current block and its first order linear estimate is encoded, it will result in certain optimal properties. Furthermore, correlation coefficient is the optimal match measure for such a coding scheme. [more]

Partially funded by LUMS Computer Science Department

  
     
 
     
 
Computer Vision LabDepartment of Computer ScienceLUMS School of Science and Engineering
LUMS, Sector U, DHA, Lahore 54792, Pakistan
All rights reserved