Varun Burde

Research

I'm interested in 3D computer vision and robotic manipulation: object-centric 3D reconstruction, 6D object pose estimation, and neural scene representations, including for transparent and textureless objects. Representative papers are highlighted.

1. Benchmarking the Effects of Object Pose Estimation and Reconstruction on Robotic Grasping Success

   V. Burde, P. Burget, T. Sattler

   ICRA 2026

   arXiv / project page / code / PerceptPick dataset

   A benchmark that ties the quality of 3D reconstruction and 6D pose estimation to what matters downstream: whether a real robot's grasp succeeds.

2. 
   OPFormer: Object Pose Estimation Leveraging Foundation Model with Geometric Encoding

   A. Moroz, V. Zeman, M. Mikšík, E. Isianova, M. David, P. Burget, V. Burde

   WACV 2026

   arXiv

   Object pose estimation that combines features from a vision foundation model with a geometric encoding of the object.

3. 
   Comparative Evaluation of 3D Reconstruction Methods for Object Pose Estimation

   V. Burde, A. Benbihi, P. Burget, T. Sattler

   WACV 2025

   arXiv / project page / code

   Which 3D reconstruction method should you use to build object models for pose estimation? A systematic comparison on the downstream task.

4. Object Pose Estimation Using Implicit Representation for Transparent Objects

   V. Burde, A. Moroz, V. Zeman, P. Burget

   ECCV-W 2024 (Spotlight)

   arXiv / Transparent-BOP data / fine-tuning data

   Implicit object representations for 6D pose estimation of transparent objects, which break most methods. Datasets released.

5. A CAD Model-Based Navigation Method for UAVs around Transmission Tower

   V. Burde, J. Soucek, A. Lazarov, P. Burget

   CASE 2024

   Using the tower's CAD model to localize and navigate a UAV around power transmission towers.

6. 
   Automatic Workspace Calibration Using Homography for Pick and Place

   V. Burde, D. S. Martinez Lema, V. Zeman, L. Kahlert, T. Jochman, P. Burget

   CASE 2023

   paper / ResearchGate

   Automatic workspace calibration with a homography, replacing tedious camera-robot calibration in pick-and-place setups.

News

• Sep 2026Started a visiting Ph.D. stay at the University of Cambridge, hosted by Dr. Olaf Wysocki in the Computer Vision for Digital Twins group.
• Mar 2026Started a visiting Ph.D. stay at ETH Zürich in the Computer Vision and Geometry Group, hosted by Prof. Marc Pollefeys and Dr. Xi Wang.
• Feb 2026Our ICRA 2026 paper on how perception quality affects grasping is out, with code and the PerceptPick dataset.
• Nov 2025OPFormer, object pose estimation with foundation-model features, is out (WACV 2026).
• Oct 2024Our paper on pose estimation for transparent objects was a spotlight at the ECCV 2024 TRICKY workshop.
• Aug 2024Chaired the Optimized UAV Systems session at IEEE CASE 2024 in Bari and presented our UAV navigation paper.
• Aug 2024Our WACV 2025 paper comparing 3D reconstruction methods for pose estimation is out.
• Aug 2023Presented automatic workspace calibration at IEEE CASE 2023 in Auckland.
• Sep 2021Started my Ph.D. at CTU Prague with Torsten Sattler and Pavel Burget.