Research Engineer · UNVEST R&D Center
Final-year PhD candidate · Koç University
I am a Research Engineer at UNVEST R&D Center and a final-year PhD candidate at Koç University, where I am an AI Fellow at the KUIS AI Center advised by Dr. Fatma Güney. My research is in computer vision with a focus on 3D reconstruction.
Before my PhD, I completed my Master's degree in Computer Science and Engineering at Koç University and my Bachelor's degree in Software Engineering at Sharif University of Technology.
A transformer-based global matching architecture for optical flow, using Optimal Transport-based flow initialization, resulting in state-of-the-art performance.
A training-free framework for quantifying uncertainty in radiance fields, applicable to any representation (e.g., NeRF and 3D Gaussian Splatting), by warping renderings across views and measuring their consistency from an unseen viewpoint.
A novel 3D Gaussian Splatting training framework that renders virtual stereo pairs and leverages their estimated depth to improve geometric accuracy and novel-view synthesis quality.
A motion-supervised framework for single-image depth estimation and multi-object segmentation, trained without ground-truth labels and supervised purely by optical flow.
A self-supervised monocular depth estimation framework that follows the traditional planar-parallax paradigm to predict metrically accurate depth, avoiding the scale ambiguity inherent in existing self-supervised approaches.
Factorizes video into static (scene structure and ego-motion) and dynamic (independently moving object) components for better stochastic prediction in driving scenes.
Jointly estimates depth and segments objects that share the same motion, from monocular video without any ground-truth labels.
Code released alongside my research.