Tomáš Hodaň
Staff Research Scientist at Meta
Zurich, Zurich, Switzerland
Summary
👤
Senior🎓
Top SchoolTomáš Hodaň is a Staff Research Scientist at Meta based in Zurich with a decade of experience in 3D computer vision, specializing in 6DoF object and hand pose estimation, tracking and reconstruction. He holds a PhD from Czech Technical University in Prague and progressed from research roles and internships at Microsoft and Google to senior research positions at Meta. Tomáš blends academic rigor with production-focused engineering—contributing to open-source tools like the BOP toolkit and BlenderProc to improve evaluation, data generation and export pipelines for photorealistic training. His work often sits at the intersection of synthetic data, benchmarking and practical deployment, reflecting a strong focus on reproducible evaluation metrics and tooling. Colleagues know him for quietly improving core infrastructure rather than flashy models, enabling more reliable real-world performance.
10 years of coding experience
19 years of employment as a software developer
PhD Computer Vision, PhD Computer Vision at Czech Technical University in PragueMaster's degree Computer Graphics and Multimedia, Master's degree Computer Graphics and Multimedia at Brno University of TechnologyMaster's degree (Erasmus) Information and Computing Sciences, Master's degree (Erasmus) Information and Computing Sciences at Utrecht University
Github Skills (8)
blender10
3d-graphics10
python10
evaluation10
rendering9
render9
command-line-arguments9
3d-reconstruction8
Programming languages (2)
Github contributions (5)
thodan/bop_toolkit
Jul 2019 - Nov 2022
A Python toolkit of the BOP benchmark for 6D object pose estimation.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:6 reviews, 87 commits, 15 PRs in 3 years 4 months
Contributions summary:Tomáš primarily focused on maintaining and improving the BOP toolkit, a Python-based project for 6D object pose estimation. They made minor changes to input/output functionalities, updated logging to align with the BOP online evaluation system, and integrated command-line arguments for setting parameters. Furthermore, the user added the average time per image calculation and implemented the BOP scores calculation. The contributions indicate a focus on the evaluation and scoring infrastructure.
object-pose-estimationpythonpose-estimationbenchmarkestimation
DLR-RM/BlenderProc
May 2020 - Jun 2020
A procedural Blender pipeline for photorealistic training image generation
Role in this project:Back-end Developer
Back-end DeveloperContributions:18 commits, 4 PRs, 8 comments in 18 days
Contributions summary:Tomáš primarily contributed to bug fixes and minor edits within the BlenderProc project, focusing on the core functionality of the application. Their work involved modifications in object loading, lighting configurations, and the sampling of rotations using the UniformSO3 provider. Furthermore, they added a new version of the BOP writer, including image data, and updated it with an exception check for images. These changes suggest a focus on improving the data generation and export capabilities of the BlenderProc pipeline.
structure-from-motionpythonblenderfeature-matching3d-engines
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.