Marco Orsingher

I am a senior algorithm engineer at VisLab (an Ambarella Inc. company), where I work on 3D reconstruction for self-driving cars. I got my PhD in computer vision from Università degli Studi di Parma, while collaborating with Ambarella Inc. research teams in Italy and Taiwan. I was also a visiting scholar at NYCU, working on knowledge distillation and neural network compression.

Previously, I was a software engineer at YAPE, designing full-stack autonomous navigation algorithms for a two-wheeled delivery robot. I hold a MSc in robotics engineering from Università degli Studi di Genova and Ecole Centrale de Nantes (double degree), as well as a BSc in control engineering from Politecnico di Milano.

Email  /  Google Scholar  /  LinkedIn  /  GitHub

I am broadly interested in computer vision and machine learning, especially in self-driving scenarios. The goal of my research is to combine geometry and learning for efficient, large-scale 3D understanding with cheap sensors.

A collection of previous works, with additional details, visualizations and experiments, as well as the proper context on how they fit into my research goals.

We analyze the sampling efficiency of NeRF and present two strategies to optimize it, given a limited training budget.

We use classical 3D reconstruction as pseudo-supervision for training NeRF and obtain much cleaner surfaces.

We show that Transformers can learn to generate point clouds with arbitrary resolution from sparse raw data.

We propose a framework for monocular 3D reconstruction that combines visual SLAM, multi-scale PatchMatch MVS with planar priors and confidence-based graph optimization.

We design a view selection algorithm that scales linearly with the number of clusters and allows to reconstruct efficiently entire cities.