Priyanka Vunnam

From Msplacements

Url: http://research.iiit.ac.in/~priyankav/resume.html
E-mail: priyankav@research.iiit.ac.in, priyanka.vunnam@gmail.com
Contact No: +91 9948618514
Download Resume: Pdf_Format [1]


Education

  • MS by Research (CSE), IIIT-Hyderabad, (expected May, 2010);
  • B. Tech. (ECE), IIIT- Hyderabad, June 2008, CGPA 8.04.
  • Senior Secondary, New Wave Junior College, Khammam, May 2004, 96.7%.

Work Experience

  • Research Assistant, CVIT, May 2008 - till date.
    Project: Symantic Analysis Methods for Visual Information
  • Teaching Assistant, Monsoon 2007
    Course: Computer Programming
  • Internship, Sarnoff Innovative Technologies, Bangalore, May - Jun 2007.
  • Teaching Assistant, Spring 2007
    Course: Computer Graphics
  • Student Intern, CVIT, May - Jul 2006.
    Project: Simulation of a Lens based Camera in OpenGL

Academic Achievements

  • Figured in the Dean's List for Academic Excellence in the Spring 2005 semester
  • Secured All India rank 1178 out of 4,00,000 applicants in AIEEE 2004.
  • Secured State rank 114 out of 2,00,000 applicants in EAMCET 2004.
  • Recipient of Pratibha scholarship award from the govt. of Andhra Pradesh.


Conferences and Workshops

  • Attended TechVista, Microsoft Research India's annual research symposium, held in Chennai, India (Oct. 2008).


Major Projects

  • Viola Jones Face Detection in the Compressed Domain
    Advisor: Dr. C.V.Jawahar
    (Honours Project, CVIT, IIIT-H.)(Team of size 2)
    Aim : Develop a compressed face detector using the compressed domain features and the Viola Jones algorithm without explicitly constructing the subsampled image.
    – Instead of constructing the subsampled (by 8) image from the DCT coefficients,these coefficients are combined to construct the integral image of the upsampled version (subsampled by 2). This integral image constructed from the compressed domain is directly used by the Viola Jones algorithm. With a little increase in the computational cost than the previous work, we achieved better accuracy.
    Technologies Used : C++, OpenCV, Dali

  • Large Scale Video Processing in Compressed Domain (Face Detection)
    Advisor: Dr. C.V.Jawahar
    (Btech Project, CVIT, IIIT-H.)(Team of size 2)
    Aim : Implement the Viola Jones face detection algorithm on compressed images by constructing a subsampled version of the original image.
    - The algorithm works around the DC and AC co-efficients present in each DCT block of the compressed image. The challenge lies in creating a subsampled(by 8) version of the original frame using compressed features and feed it to the Viola Jones detection. We studied the performance of this algorithm by running it on datasets available in the public domain. One such dataset is the Yale dataset. For this dataset our results were more accurate than the traditional VJ detector and the computational time significantly reduced.
    Technologies Used : C++, OpenCV, Dali

  • Video Retrieval in Compressed Domain
    Advisor: Dr. C.V.Jawahar
    (Btech Project, CVIT, IIIT-H.)(Team of size 2)
    Aim : Build a video retrieval system on news videos with feature extraction done in the compressed domain which significantly improves computational costs.
    - This project mainly includes feature extraction from videos and using them for indexing, comparison and thereby, retrieval from large databases. We worked on news videos and built a complete system which shows top 10 similar videos to the query video. The features we implemented are color, texture and motion activity. There three features are very straightforward in the pixel domain. Implemented the optimistic estimates of these three features in the compressed domain and used this feature vector for indexing the videos from huge database.
    Technologies Used: C++, Dali, PHP, MySQL

  • 3D Photo Browser
    Advisor: Dr. C.V.Jawahar
    (Semester Project for the course Computer Vision)(Team of size 16)
    Aim : Developing a user interface which allows the user to upload photographs, give manual assistance in 3D model building, interfaces for saving photograph/3D model based collections that can be reloaded, and finally controls for moving around in the 3D world.
    - It had 4 modules: a)Fundamental matrix computation, b)Internal(Intrinsic) parameter computation, c)3D reconstruction and Bundle adjustment and d)User Interface. The third module was developed by our team which had 4 members and which deals with computation of the 3D point cloud, given correspondences across cameras along with camera intrinsics/extrinsics. Bundle adjustment for improving the solutions provided by the above modules so as to increase the accuracy of reconstruction.
    Technology Used: C++, Matlab

  • MPEG Video Processing in the Compressed Domain
    Advisor: Dr. C.V.Jawahar
    (Honours Project, CVIT, IIIT)(Team of size 1)
    Aim : Apply simple pixel-domain image processing techniques to videos directly in the compressed state.
    – Since the multimedia content is stored and transferred in the compressed format, the data needs to be fully decompressed before applying any pixel-domain image processing techniques which is computationally expensive for videos. Again the processed data needs to be compressed for easy storage and transferring. To avoid this decompression-recompression step, the video data can be processed in the compressed domain directly. The compressed videos are decoded to the semi-compressed level (DCT level), applied the desired functions (dct-domain techniques equivalent to pixel-domain techniques) and then compress the processed data. Implemented the functions zooming and croping of mpeg videos in the compressed domain.
    Technologies Used: C++, Dali

Other Projects

  • Support Cluster Machine and study SVM libraries done as part of Machine Learning course ( Monsoon 2008 )
  • 3D photo browser done as part of Computer vision ( Spring 2007 )
  • Image Registration using Sub-pixel Alignment done as part of Digital Image Processing course ( Monsoon 2006 )
  • BCH Error Correcting codes done as part of Information Theory and Coding ( Monsoon 2006 )
  • Developed IGL Library which aims at developing and implementing various GL functions as part of Computer Graphics ( Spring 2006 )
  • Built Tetris Game and Simulated Train done as part of Computer Graphics ( Spring 2006 )
  • Built a digital timer as part of Electronics Workshop course ( Monsoon 2005)


Skillset

  • Programming Languages  : - C, C++
  • Operating Systems  : - GNU/Linux, Windows 95/98/2000/XP.
  • Scripting Languages  : - Python, Matlab.
  • Web Languages  : - HTML, XML, PHP.
  • Libraries  : - Qt, OpenGL, OpenCV
  • Tools  : - MySQL, MS Office
  • Publishing  : - LaTeX

Advanced Courses Taken

Machine Learning, Pattern Recognition, Computer Vision, Computer Graphics, Digital Image Processing, Digital Signal Processing, Information Theory and Coding, Cryptography and Network Secutiry, Communication Networks, Parallel Programming, Linear Programming and Advanced Linear Programming

Academic Interests

Machine Learning, Pattern Recognition, Computer Vision, Image Processing, Information Extraction and Retrieval, Algorithms, Programming

Extra Curriculur

  • Won many prizes in various IIIT inter-house competitions like athletics, long jump.
  • Successfully organized the event "Chat A Pic" in Felicity 07.

Other Interests

  • Extra-academic: Enjoy studying philosophy.
  • Sports: Athletics, Shuttle, Carroms
  • Others: Sketching, Travelling
Retrieved from "http://www.iiit.ac.in/msplacements/index.php/Priyanka_Vunnam"
Personal tools