Dr. rer. nat. Eric Brachmann

Office: 1331, Appelstr. 9A, D-30167 Hannover, Germany
E-Mail: name.surname@tu-dresden.de
Phone: +49 511 762 5322
Twitter/News: @eric_brachmann

Research Interests - Research Experience - Awards - Industry Experience - Teaching - Publications

Research Interests:

Computer Vision, Machine Learning, Object Pose Estimation, Camera Localization

Research Experience:

Since 2019

Guest at the Leibniz University Hannover in the group of Prof. Bodo Rosenhahn


Doctorate awarded by the TU Dresden (Dr. rer. nat., awarded with Summa Cum Laude)

Since 2017

Research Associate in the Visual Learning Lab of Prof. Rother at the University Heidelberg


Guest at the Center for Systems Biology Dresden in the group of Florian Jug


Research visit, Microsoft Research Cambridge (3 months, host: Sebastian Nowozin)

2012 – 2017

Research associate and PhD student at the TU Dresden,
partly Computer Graphics and Vizualisation Lab of Prof. Gumhold,
partly Computer Vision Lab Dresden of Prof. Rother


Diplom in media computer science, passed with distinction

2006 – 2012

Studies of media and computer science at TU Dresden

Reviewing Activities:

  • CVPR 18, 19, 20; ICCV 19; ECCV 18; NeurIPS 19; TPAMI 19, 20; IJCV 18; JMLR 19; ICRA 18; IROS 17, 18; GCPR 15, 17, 18; ICCV Workshops 17
  • Outstanding Reviewer: CVPR 19; NeurIPS 19 (Top 400)

Workshops / Tutorials:

  • Co-Organizer of Large-Scale Visual Place Recognition and Image-Based Localization,
    ICCV Tutorial, 2019 (link)
  • Co-Organizer of 5th International Workshop on Recovering 6D Object Pose (R6D),
    ICCV Workshop, 2019 (link)
  • Co-Organizer of Visual Localization: Feature-based vs. Learned Approaches,
    ECCV Tutorial, 2018 (link)
  • Invited Speaker at Workshop on Geometry Meets Deep Learning,
    ECCV Workshop, 2018 (link)
  • Invited Speaker at Workshop on Learnable Representations for Geometric Matching,
    CIIRC Prague, 2017



Nominated for GI Dissertation Award 2018 by the TU Dresden
(each university in Germany, Austria and Switzerland nominates one computer science dissertation for the award, annually)


ACCV Honorable Mention Demo Award
(for our paper: Learning Analysis-by-Synthesis for 6D Pose Estimation in RGB-D Images)


Enno Heidebroek Award
(awarded to the best graduates of the engineering department of the TU Dresden)

2008 – 2012

Scholarship of the German National Academic Foundation
(awarded to students with exceptional academic performance, extracurricular interests and social commitment)


IBM Award
(awarded to students with an exceptional intermediate diploma)

Industry Experience:

11/2010 – 03/2011

Internship at IBM Deutschland Research and Development GmbH

09/2008 – 04/2009

Student employee at T-Systems Multimedia Solutions GmbH

04/2008 – 08/2008

Internship at T-Systems Multimedia Solutions GmbH

Teaching Experience:

Preparation of lectures for Computer Vision I (Prof. Rother, TU Dresden, 2015-2017), Reconstructing and Understanding the 3D World (Prof. Rother, Heidelberg University, 2018); organizing and conducting exercises accompanying Computer Graphics I (Prof. Gumhold, TU Dresden, 2013-2017), supervisor of numerous Diploma; Master and Bachelor theses with focus on computer vision and machine learning; conducting practical courses and seminars with focus on computer vision and robotics


Reinforced Feature Points: Optimizing Feature Detection and Description for a High-Level Task
tldr: "end-to-end learning of sparse key points in a complete vision pipeline including discrete matching and robust model fitting"
Aritra Bhowmik, Stefan Gumhold, Carsten Rother, Eric Brachmann
CVPR 2020 (oral, paper)

CONSAC: Robust Multi-Model Fitting by Conditional Sample Consensus
tldr: "extension of NG-RANSAC to find multiple instances of a parametric model by learned sequential search"
Florian Kluger, Eric Brachmann, Hanno Ackermann, Carsten Rother, Michael Yang, Bodo Rosenhahn
CVPR 2020 (papercode)

Expert Sample Consensus Applied to Camera Re-Localization
tldr: "ESAC, a combination of mixture-of-experts and RANSAC to fit a parametric model in large and ambiguous domains"
Eric Brachmann,
 Carsten Rother
ICCV 2019 (paperproject page)

Neural-Guided RANSAC: Learning Where to Sample Model Hypothesis
tldr: “NG-RANSAC, a neural network guides random sampling of parametric model candidates”
Eric Brachmann, Carsten Rother
ICCV 2019 (paperproject page)

iPose: Instance-Aware 6D Pose Estimation of Partly Occluded Objects
tldr: “position and orientation of objects by instance segmentation and deep object coordinate prediction”
Omid Hosseini Jafari, Siva Karthik Mustikovela, Karl Pertsch, Eric Brachmann, Carsten Rother
ACCV 2018 (paper)

BOP: Benchmark for 6D Object Pose Estimation
tldr: “evaluation of RGB-D pose estimation methods, learned and not learned, on diverse datasets”
Tomas Hodan, Frank Michel, Eric Brachmann, Wadim Kehl, Anders Buch, Dirk Kraft, Bertram Drost, Joel Vidal, Stephan Ihrke , Xenophon Zabulis, Caner Sahin, Fabian Manhardt, Federico Tombari, Tae-Kyun Kim, Jiri Matas, Carsten Rother
ECCV 2018 (paper, project page)

Learning to Predict Dense Correspondences for 6D Pose Estimation
tldr: “summary of my work prior to 2018, learning object and scene coordinate regression using random forests and neural networks”
Eric Brachmann
PhD Thesis (online version)

Learning Less is More - 6D Camera Localization via 3D Surface Regression
tldr: “DSAC++, an improved version of DSAC (CVPR17), learning scene coordinate regression without a 3D model or depth, and with differentiable PnP”
Eric Brachmann, Carsten Rother
CVPR 2018 (paperproject page)

DSAC - Differentiable RANSAC for Camera Localization
tldr:  “learning a neural network such that its predictions work well with RANSAC-based model fitting”
Eric Brachmann, Alexander Krull, Sebastian Nowozin, Jamie Shotton, Frank Michel, Stefan Gumhold, Carsten Rother
CVPR 2017 (oral, paperproject page)

Global Hypothesis Generation for 6D Object Pose Estimation
tldr: “instead of RANSAC-based sampling to find pose inliers, optimize energy in a graphical model”
Frank Michel, Alexander Kirillov, Eric Brachmann, Alexander Krull, Stefan Gumhold, Bogdan Savchynskyy, Carsten Rother
CVPR 2017 (spotlight, paperproject page)

PoseAgent: Budget-Constrained 6D Object Pose Estimation via Reinforcement Learning
tldr: “an RL agent iteratively selects object pose hypotheses for refinement”
Alexander Krull, Eric Brachmann, Sebastian Nowozin, Frank Michel, Jamie Shotton, Carsten Rother
CVPR 2017 (paperproject page)

Random Forests versus Neural Networks - What's Best for Camera Relocalization?
tldr:  “investigates mapping of random forests to NNs for optimization, and back again for efficiency”
Daniela Massiceti, Alexander Krull, Eric Brachmann, Carsten Rother, Philip H.S. Torr
ICRA 2017  (paper)

Uncertainty-Driven 6D Pose Estimation of Objects and Scenes from a Single RGB Image
tldr: “estimate poses of multiple objects or scenes from RGB, predict object coordinate distributions and search for max likelihood pose”
Eric Brachmann, Frank Michel, Alexander Krull, Michael Ying Yang, Stefan Gumhold, Carsten Rother
CVPR 2016 (papersupplementproject page)

Learning Analysis-by-Synthesis for 6D Pose Estimation in RGB-D Images
tldr: “substitute inlier counting pose score with a CNN that compares input image and renderings, trained via max likelihood”
Alexander Krull, Eric Brachmann, Frank Michel, Michael Ying Yang, Stefan Gumhold, Carsten Rother
ICCV 2015 (paperproject page)

Pose Estimation of Kinematic Chain Instances via Object Coordinate Regression
tldr:  “pose estimation of articulated objects, only n+2 points are needed to estimate n-jointed objects”
Frank Michel, Alexander Krull, Eric Brachmann, Michael Ying Yang, Stefan Gumhold, Carsten Rother
BMVC 2015 (oral, papersupplementproject page)

6-DOF Model Based Tracking via Object Coordinate Regression
tldr: “combine RANSAC-based hypothesis sampling with a particle filter for real-time tracking of position and orientation”
Alexander Krull, Frank Michel, Eric Brachmann, Stefan Gumhold, Stephan Ihrke, Carsten Rother
ACCV 2014 (oral, Honorable Mention Demo Award, papersupplementproject page)

Learning 6D Object Pose Estimation using 3D Object Coordinates
tldr: “predict dense image-to-object correspondences with a random forest, and solve for 6D pose with RANSAC”
Eric Brachmann, Alexander Krull, Frank Michel, Stefan Gumhold, Jamie Shotton, Carsten Rother
ECCV 2014 (papersupplementproject page)

Feature propagation on image webs for enhanced image retrieval
tldr:  “propagate visual words along image web edges to make a BoW image descriptors more robust”
Eric Brachmann, Marcel Spehr, Stefan Gumhold
ICMR 2013 (oral, paper)

Simplified Authentication and Authorization for RESTful Services in Trusted Environments
tldr: “an authentication scheme for company intranets where you may want to trade security for simplicity”
Eric Brachmann, Gero Dittmann, Klaus-Dieter Schubert
ESOCC 2012