About Me
Currently I am a Ph.D. candidate, interested in computer vision and specifically non-parametric visual object tracking, at Sabanci University Electronics Engineering 
and a research assistant at VPALAB working on TUBITAK project named "Fast and Convergent Algorithms for Large Scale Learning".
Before my Ph.D. studies, I received my B.Sc. and M.Sc. degrees from Yildiz Technical University Mathematical Engineering at 2004 and 2008 respectively.
If you are interested in my resume, you can check my LinkedIn Profile . Also you can contact me through my e-mail isaygint [at] sabanciuniv [dot] edu.
Photo taken at Niagara Falls, NY, USA at September 2011
Object Tracking
My Ph.D. research is focused on visual object tracking.
Related Publications
- Topkaya, I.S.; Erdogan, H.; , "Confidence-Based Dynamic Classifier Combination For Mean-Shift Tracking", (Submitted to) Pattern Recognition Letters 
- Topkaya, I.S.; Erdogan, H.; , "Histogram correlation based classifier fusion for object tracking", IEEE 19th Conference on Signal Processing and Communications Applications (SIU), 2011
Large Scale Learning
Currently I am involved in "Fast and Convergent Algorithms for Large Scale Learning" project which is supported by The Scientific and Technological Research Council of Turkey (TUBITAK) .
Our main objective is to develop QMM library (a LIBLINEAR mod) which is a general solver for binary and multi-class regularized empirical risk classification.
Audio-Visual Speech Recognition
Between September 2008 and September 2010, I have worked in "Novel Approaches in Audio-Visual Speech Recognition" project which was supported by The Scientific and Technological Research Council of Turkey (TUBITAK) .
During the project we have investigated effects of incorporating visual information into multi stream hidden Markov models and specifically improving recognition accuracy with tandem classifiers and coupled hidden Markov models.
We have also participated in eNTERFACE '10 workshop with a live audio-visual speech recognition application, report of which can be found in workshop proceedings.
Related Publications
- Topkaya, I.S.; Erdogan, H.; , "Using multiple visual tandem streams in audio-visual speech recognition", IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), 2011
- Topkaya, I.S.; Sen, M.U.; Yilmaz, M.B.; Erdogan, H.; , "Improving speech recognition with audio-visual tandem classifiers and their fusions", IEEE 19th Conference on Signal Processing and Communications Applications (SIU), 2011
- Topkaya, I.S.; Yilmaz, M.B.; Sen, M.U.; Tarasov, A.; Erdogan, H.; , "An Audio-Visual Speech Recognition System with Live Inputs", 6th International Summer Workshop on Multimodal Interfaces (eNTERFACE), 2010
Face Recognition
During my Master's study at Yildiz Technical University , I have worked on face recognition from videos. My master thesis study was on a face recognition system that selects and processes frames from videos to improve recognition accuracy. The work can be seen in the ISCIS '08 conference paper.
Related Publications
- Topkaya, I.S.; Bayazit, N.G.; , "Improving face recognition from videos with preprocessed representative faces", 23rd International Symposium on Computer and Information Sciences (ISCIS), 2008