NeuraVision · Computer Vision & Deep Learning · Bilkent
We study the topology of what neural networks see.
We build computer-vision and deep-learning methods that respect the geometry and topology of the visual world — and that know when they might be wrong.
The Lab
About the lab
NeuraVision is a computer-vision and machine-learning research group in the Department of Computer Engineering at Bilkent University. Led by Dr. Doruk Öner, the lab develops deep-learning methods that capture the geometry and topology of images — from delineating thin, connected structures like roads, neurons and blood vessels, to estimating predictive uncertainty for trustworthy AI. Our work spans medical image analysis, remote sensing, 3D shape modeling, and the theory of structured prediction, and has appeared in venues such as TPAMI, ICML, MICCAI and TMLR.
Research
What we work on
Several threads, one question: how can deep networks respect the structure, geometry and uncertainty of the visual world?
Topological Deep Learning
Loss functions and representations that make neural networks respect the connectivity and topology of what they predict.
Explore → 02 · VisionCurvilinear Structure Delineation
Detecting thin, connected networks — roads, blood vessels, neurons, cracks — where a single missed pixel breaks the whole graph.
Explore → 03 · TrustUncertainty Estimation & Trustworthy AI
Models that know when they might be wrong — calibrated, efficient uncertainty for high-stakes decisions.
Explore → 04 · HealthMedical Image Analysis
Bringing structure-aware and uncertainty-aware deep learning to clinical imaging.
Explore → 05 · 3D3D Shape & Implicit Representations
Part-based implicit neural representations for composite 3D shapes and optimization.
Explore → 06 · Motion ForecastingInteraction-Aware Motion Forecasting
Forecasting agent motion in autonomous-driving scenes by modeling interactions, uncertainty and multimodal future behavior.
Explore →Selected Work
Featured publication
Latest
Lab news
- Sep 2025 CAPE accepted at MICCAI 2025New Lab members Elyar Esmaeilzadeh, Ehsan Garaaghaji and Farzad Hallaji Azad, with Dr. Doruk Öner, present CAPE — a Connectivity-Aware Path Enforcement loss for curvilinear structure delineation — at MICCAI 2025.
- Feb 2025 Detecting bronchiolitis obliterans from chest CT Our collaborative work harnessing deep learning to detect bronchiolitis obliterans syndrome from chest CT appears in Communications Medicine.
- Jan 2025 PartSDF published in TMLR PartSDF, a part-based implicit neural representation for composite 3D shape parametrization and optimization, is published in Transactions on Machine Learning Research.
- Sep 2024 NeuraVision Research Lab established at Bilkent University Dr. Doruk Öner founds the NeuraVision Research Lab in the Department of Computer Engineering at Bilkent University, focusing on topology-aware computer vision and trustworthy deep learning.
- Jul 2024 Uncertainty estimation in iterative neural networks at ICML 2024 Our method for enabling efficient uncertainty estimation in iterative neural networks is presented at ICML 2024.
Join the lab.
We are looking for curious students who like hard, beautiful problems in vision and learning.