My research interests include but not limited to visual representation learning (deep metric learning), content based image retrieval, deep model reduction via knowledge distillation, learned image compression, recognition and (weakly-supervised) detection, semantic segmentation, multi-target tracking, and stereo matching.

For my full publication list, please visit my Google Scholar Page. My Ph.D. thesis is about developing mathematical models to explain and to improve the practices in deep metric learning (lib.metu) and my M.Sc. thesis is about building dynamic programing models for fast stereo matching (lib.metu).