NeuraVision Research lab
The Neuravision lab led by Doruk Oner specializes in computer vision and deep learning. Our main goal is to teach deep models about topology. We investigate diverse applications of topological deep learning in segmentation of curvilinear structures in medical imaging, such as blood vessels and neurons, and also in satellite imagery such as road networks. Additionally, we are exploring research topics such as 3D implicit representation, trajectory and motion forecasting and uncertainty estimation.
Research
What we work on
Several threads, one question: how can deep networks respect the structure, geometry and uncertainty of the visual world?
01 · Topology
Topology-Aware Computer Vision
We develop computer vision methods that explicitly account for topological properties such as connectivity, continuity, and the preservation of meaningful structures. Our work focuses on topology-aware learning for curvilinear and network-like structures in both 2D and 3D, with applications including vessels, neuronal structures, road networks, and other connected visual patterns.
Explore →
02 · Trust
Uncertainty Estimation & Trustworthy AI
We develop methods for estimating predictive uncertainty and improving the reliability and trustworthiness of deep learning models. Our research focuses on calibrated confidence, uncertainty-aware prediction, robustness, and identifying unreliable outputs, particularly in settings where dependable model behavior is critical.
Explore →
03 · 3D
3D Shape & Implicit Representations
We study neural representations for modeling, understanding, and manipulating complex 3D shapes. Our work includes implicit representations and part-based approaches for shape reconstruction, generation, parametrization, editing, and optimization, with an emphasis on representing geometry in flexible and controllable ways.
Explore →
04 · Motion Forecasting
Trajectory & Motion Forecasting
We develop methods for predicting the future trajectories and motion of agents in dynamic environments. Our work spans human trajectory forecasting and autonomous-driving scenarios, with a focus on multimodal prediction, interactions between agents, and modeling plausible future behavior in complex scenes.
Explore →Selected Work
Latest publications
-
ACCV · 2026
Direction-Aware Kinematics-Conditioned Graph Hypernetwork for Human Trajectory Prediction
-
BMVC · 2026
Stochastic Nonlinearities Improve Uncertainty Estimation
-
SIU · 2026
Track-Aligned Interaction Modeling for Group Activity Recognition
-
MICCAI · 2025
CAPE: Connectivity-Aware Path Enforcement Loss for Curvilinear Structure Delineation
@inproceedings{esmaeilzadeh2025cape,
title = {{CAPE: Connectivity-Aware Path Enforcement Loss for Curvilinear Structure Delineation}},
author = {Esmaeilzadeh, Elyar and Garaaghaji, Ehsan and Hallaji Azad, Farzad and {\"{O}}ner, Doruk},
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025},
pages = {183--193},
year = {2025},
publisher = {Springer Nature Switzerland},
doi = {10.1007/978-3-032-05162-2_18}
}
-
TMLR · 2025
PartSDF: Part-based implicit neural representation for composite 3D shape parametrization and optimization
@article{talabot2025partsdf,
title = {{PartSDF: Part-Based Implicit Neural Representation for Composite 3D Shape Parametrization and Optimization}},
author = {Talabot, Nicolas and Clerc, Olivier and Demirtas, Arda Cinar and Goujon, Alexis and Le, Hieu and {\"{O}}ner, Doruk and Fua, Pascal},
journal = {Transactions on Machine Learning Research},
issn = {2835-8856},
year = {2025},
url = {https://openreview.net/forum?id=zl43C1yBKv}
}
-
MVA · 2025
Vision-based power line cables and pylons detection for low flying aircraft
@article{gwizdala2025vision,
title = {{Vision-based power line cables and pylons detection for low flying aircraft}},
author = {Gwizda{\l}a, Jakub and {\"{O}}ner, Doruk and Roy, Soumava Kumar and Shah, Mian Akbar and Eberhard, Ad and Egorov, Ivan and Kr{\"{u}}si, Philipp and Yakushev, Grigory and Fua, Pascal},
journal = {Machine Vision and Applications},
volume = {36},
number = {2},
pages = {46},
year = {2025},
doi = {10.1007/s00138-025-01664-1}
}
Latest
Lab news
- Sep 2026 Direction-Aware Kinematics-Conditioned Graph Hypernetwork for Human Trajectory PredictionNew Amirreza M. Shebly, Mohammad Sarhangzadeh, and Doruk Öner have had their paper accepted to ACCV 2026.
- Aug 2026 Stochastic Nonlinearities Improve Uncertainty Estimation Mohammad Sarhangzadeh, Amirreza M. Shebly, and Doruk Öner have had their paper accepted to BMVC 2026.
- Jul 2026 Track-Aligned Interaction Modeling for Group Activity Recognition Amirreza M. Shebly and Doruk Öner have had their paper accepted to SIU 2026.
- Sep 2025 CAPE accepted at MICCAI 2025 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.
Join the lab.
We are looking for curious students who like hard, beautiful problems in vision and learning.