PeopleofSuperintelligence

Researcher

Federico Tombari

Federico Tombari is a computer vision researcher whose work connects 3D perception, machine learning and multimodal AI. In 2009, he completed a PhD at the University of Bologna on visual correspondence; in 2012, a Hough-voting paper with Luigi Di Stefano recognized 3D shapes under occlusion and clutter. CNN-SLAM fused learned depth with monocular SLAM in 2017, and he co-founded Pointu3D, a Munich startup for 3D perception in AR and robotics, in 2018–2019; Google acquired it in 2019. SceneGraphFusion built semantic 3D scene graphs in 2021, SceneFun3D introduced interaction annotations for indoor 3D scenes in 2024, and he coauthored TokenFormer in 2025, which treated model parameters as tokens and scaled by adding key-value parameters. As of October 2026, he is a Research Director at Google Zurich, where he leads a team on 3D computer vision and multimodal machine learning for XR products, and a lecturer (PrivatDozent) at TUM.

Works
509
Citations
22,192
h-index
68
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2009

    A doctorate in visual correspondence

    In 2009, he completes a PhD at the University of Bologna with a dissertation on visual correspondence, including fast pattern matching, stereo correspondence and robust image matching.

  2. 2012

    Hough voting for 3D shapes

    In 2012, he and Luigi Di Stefano publish a method that matches 3D features and uses their consensus in a 3D Hough space to recognize shapes under occlusion and clutter.

  3. 2017

    CNN-SLAM with learned depth

    In 2017, he coauthors CNN-SLAM with Keisuke Tateno, Iro Laina and Nassir Navab. It fuses CNN-predicted dense depth with direct monocular SLAM and combines semantic labels from a single frame with dense SLAM to make a semantically coherent scene reconstruction from one view.

  4. 2018–2019

    Co-founding Pointu3D

    In 2018–2019, he co-founds Pointu3D and serves as its managing director; the Munich startup works on 3D perception for AR and robotics. Google acquires Pointu3D in 2019.

  5. 2021

    SceneGraphFusion for 3D scenes

    In 2021, he coauthors SceneGraphFusion, which builds semantic scene graphs incrementally from RGB-D frames by aggregating PointNet features with a graph neural network.

  6. 2024

    A dataset for 3D interaction

    In 2024, he coauthors SceneFun3D, a dataset with more than 14,800 interaction annotations across 710 high-resolution indoor 3D scenes. It supports functionality segmentation, task-driven affordance grounding and 3D motion estimation.

  7. 2025

    Scaling Transformers with TokenFormer

    In 2025, he coauthors TokenFormer, which treats model parameters as tokens and replaces Transformer linear projections with attention in which input tokens query parameter keys and values. The model scales by incrementally adding key-value parameters.

  8. As of October 2026

    Research Director at Google Zurich

    As of October 2026, he is a Research Director at Google Zurich, where he leads a research and product team on 3D computer vision and multimodal machine learning for XR products. He is also a lecturer (PrivatDozent) at TUM’s Faculty of Computer Science, affiliated with the CAMP Chair.

Career

  1. Since 2019GoogleResearch Director
  2. Since 2015Technical University of MunichLecturer (PrivatDozent)
  3. 2018–2019Pointu3DManaging Director

Education

  1. 2009University of BolognaPhD

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

  1. 11% of AI doctorate holders from 2010 to 2016 founded or led a company or lab by 10 years after graduationIncludes Federico Tombari. We estimate how often AI doctorate holders founded or led a company or lab by 10 years after graduation, how waiting times compare between founders and top jobs, and which universities trained the most.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  4. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Federico Tombari yet. Send us a free-licensed photo.

  1. Methodologies for visual correspondence nbn.depositolegale.it. Checked 22 October 2026.
  2. Hough Voting for 3D Object Recognition under Occlusion and Clutter jstage.jst.go.jp. Checked 22 October 2026.
  3. CNN-SLAM: Real-Time Dense Monocular SLAM With Learned Depth Prediction openaccess.thecvf.com. Checked 22 October 2026.
  4. Federico Tombari research.google. Checked 22 October 2026.
  5. Highlights - Federico Tombari federicotombari.github.io. Checked 22 October 2026.
  6. SceneGraphFusion: Incremental 3D Scene Graph Prediction From RGB-D Sequences openaccess.thecvf.com. Checked 22 October 2026.
  7. SceneFun3D: Fine-Grained Functionality and Affordance Understanding in 3D Scenes openaccess.thecvf.com. Checked 22 October 2026.
  8. Federico Tombari - Lehrstuhl für Informatikanwendungen in der Medizin cs.cit.tum.de. Checked 22 October 2026.
  9. Machines Are Learning to Understand Space — Federico Tombari at the AI Symposium on 3D AI, Explainability, and Immersive Worlds hun-ren.hu. Checked 22 October 2026.