I am a third-year PhD student in the field of computer vision and machine learning, supervised by Prof. Theo Gevers and co-supervised by Dr. Sezer Karaoğlu at the Delta Lab, which is a collaboration between the University of Amsterdam and Bosch. I closely collaborate with Dr. Maxim Tatarchenko and Dr. Vien Ngo. My research interests mainly lie in 3D generative models, scene understanding, 3D reconstruction and view synthesis for augmented reality and robotic applications.
GaussianTeller is a native 3D Gaussian splat generation framework that directly learns 3D diffusion priors from spatially-grouped Gaussians encoded into a structured latent space.
Built using in-context learning, CAD model retrieval and 3DGS-based stylization, SceneTeller generates realistic and high-quality 3D spaces from natural language prompts.
RealDiff formulates point cloud completion as a conditional generation problem directly on real-world measurements in a self-supervised way. To better deal with noisy observations, we leverage additional geometric cues.
A lateral vehicle control network can be trained from only an unlabeled sequence of images using novel view synthesis, without the need for a specialized setup on the car.