Andreas Geiger
This website maintains a list of older projects and is not updated anymore. For our more recent projects, visit:

Research Projects

  • Robust Vision Challenge

    Competition at CVPR 2018 across various computer vision tasks.
  • DALI 2018

    DALI 2018 Workshop on Autonomous Driving Talks.
  • ETH3D Benchmark

    Multi-view 3D reconstruction benchmark and evaluation.
  • SlowFlow

    Exploiting high-speed cameras for optical flow reference data.
  • OctNetFusion

    Learning coarse-to-fine depth map fusion from data.
  • OctNet

    Learning 3D representations at high resolutions with octrees.
  • Autonomous Vision

    Survey article on research in vision for autonomous vehicles.
  • Similarity Priors

    Joint detection and volumetric 3D reconstruction of similar objects.
  • 3D Reconstruction

    Probabilistic volumetric 3D reconstruction with ray potentials.
  • FollowMe

    Computationally Efficient Online Min-Cost Flow Tracking.
  • Discrete Flow

    Optical flow estimation cast as inference in a discrete CRF.

    Efficient MAP inference with sparse high-order potentials.
  • Indoor Scenes

    3D layout and 3D object inference from a single RGB-D image.
  • Semi-Convex Hull

    Reduces a complex CAD model to a watertight low-face mesh.
  • Displets

    Resolving stereo ambiguities using knowledge about objects.
  • Object Scene Flow

    Scene flow for autonomous vehicles. Novel realistic dataset.

    Lightweight MATLAB tool for semantic and instance labeling.

    OpenGl rendering of 3D objects directly from within MATLAB.
  • Omnidirectional Vision

    Catadioptric camera calibration software and dataset.
  • Tracking by Detection

    Simple hierarchical object tracker using the DPM detector.
  • TriTrack

    Real-time object tracking using sparse scene flow features.
  • 3D Urban Scenes

    Code and datasets for 3D urban traffic scene understanding.
  • KITTI Vision Suite

    Datasets and benchmarks for computer vision and robotics.
  • LOST

    Localization in OpenStreetMap data using visual odometry only.

    MATLAB code for automatic corner / checkerboard detection.
  • Karlsruhe Objects

    Images from a moving vehicle with annotated cars / pedestrians.
  • Karlsruhe Sequences

    Urban stereo video sequences recorded from a moving vehicle.

    MATLAB / C++ code for efficent large-scale stereo matching.

    MATLAB / C++ code for stereo and monocular visual odometry.

    MATLAB / C++ code for iterative closest point matching (2D/3D).
  • Gaussian Processes

    Java Applet illustrating Gaussian Process regression.
  • Multi-Cam Calibration

    Geometric and photometric multi-camera calibration.

eXTReMe Tracker