Home
Uni-Logo
 

Binaries/Code

Binaries/Code Datasets Open Source Software

We provide binaries and source code of some selected works in order to help other researchers to compare their results or to use our work as a module for their research. Please understand that we can only provide what is offered here. E-Mails requesting other free code will be ignored.

Terms of use

All code is provided for research purposes only and without any warranty. Any commercial use requires our consent. When using the code in your research work, you should cite the respective paper. Refer to the readme file in each package to learn how to use the program.


DispNet: A Convolutional Neural Network for Estimating Disparities

Download modified master branch Caffe with model included: Download v0.5 (Modified Caffe + models + brief readme)
Note: This is release v0.5 which does not include training yet. Code for training will be added very shortly.
Please read the included DISPNET-README.md file.


N. Mayer, E. Ilg, P. Häusser, P. Fischer, D. Cremers, A. Dosovitskiy, T. Brox
A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation,
IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2016.

Generating Images with Perceptual Similarity Metrics based on Deep Networks

Download      Trained models for layers pool5-fc8 and a python demo        Trained models for layers norm1-conv4

A. Dosovitskiy, T. Brox
Generating Images with Perceptual Similarity Metrics based on Deep Networks,
arXiv pre-print, 2016.

Motion Trajectory Segmentation via Minimum Cost Multicuts

Download Executables for 64-bit Linux

M. Keuper, B. Andres, T. Brox
Motion Trajectory Segmentation via Minimum Cost Multicuts,
IEEE International Conference on Computer Vision (ICCV), 2015.

Global, Dense Multiscale Reconstruction for a Billion Points

Download Executables for 64-bit Linux

Project page

B. Ummenhofer, T. Brox
Global, Dense Multiscale Reconstruction for a Billion Points,
IEEE International Conference on Computer Vision (ICCV), 2015.

FlowNet: Learning Optical Flow with Convolutional Networks (v1.0)

Download modified master branch Caffe with model included: Download v1.0 (Modified Caffe + models + brief readme)
Note: This is release v1.0 which includes both FlowNetS and FlowNetC.
Please read the included FLOWNET-README.md file.

Older versions: (v0.1)

A. Dosovitskiy and P. Fischer and E. Ilg and P. Häusser and C. Hazirbas and V. Golkov and P. v.d. Smagt and D. Cremers and T. Brox
FlowNet: Learning Optical Flow with Convolutional Networks,
IEEE International Conference on Computer Vision (ICCV), 2015.

Inverting Visual Representations with Convolutional Networks

Code and examples for new Caffe version (2016): training config and a trained model for reconstruction from FC6
To be used with this Caffe https://github.com/dosovits/caffe-fr-chairs
Download modified Caffe used in the paper: modified Caffe + brief readme
Download trained models from the paper: conv1   conv2   conv3   conv4   conv5   fc6   fc7   fc8   caffenet


A. Dosovitskiy and T. Brox
Inverting Visual Representations with Convolutional Networks,
IEEE Conference in Computer Vision and Pattern Recognition (CVPR), 2016.

Learning to Generate Chairs with Convolutional Neural Networks

New version in new Caffe (2016) trained model, demo, training example
To be used with this Caffe https://github.com/dosovits/caffe-fr-chairs
Version used for the CVPR paper modified Caffe + models + matlab scripts        trained model        training data

A. Dosovitskiy, J. T. Springenberg and T. Brox
Learning to Generate Chairs with Convolutional Neural Networks,
IEEE Conference in Computer Vision and Pattern Recognition (CVPR), 2015.


Scene flow from RGB-D sequences

Download C++ Code

J. Quiroga, Thomas Brox, F. Devernay, J. Crowley
Dense semi-rigid scene flow estimation from RGBD images,
European Conference on Computer Vision (ECCV), 2014.

Exemplar Convolutional Neural Networks

Download Code for Linux

A. Dosovitskiy, J. T. Springenberg, M. Riedmiller and T. Brox
Discriminative Unsupervised Feature Learning with Convolutional Neural Networks,
Advances in Neural Information Processing Systems 27 (NIPS), 2014.

A. Dosovitskiy, P.Fischer, J. T. Springenberg, M. Riedmiller and T. Brox
Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks,
IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2015.


Image Descriptors based on Curvature Histograms

Download Code for Linux (contains code for combination with HOG and SIFT)

The download provides feature computation code for integration with the Felzenszwalb DPM code and for integration with the VLfeat framework.

P. Fischer, T. Brox
Image Descriptors based on Curvature Histograms,
German Conference on Pattern Recognition (GCPR), 2014.


Point-Based Reconstruction

Download Executable for 64-bit Linux (requires CUDA 5.5)

B. Ummenhofer, T. Brox
Point-Based 3D Reconstruction of Thin Objects,
IEEE International Conference on Computer Vision (ICCV), 2013.


Non-smooth Non-convex Optimization

Download Matlab Code

P. Ochs, A. Dosovitskiy, T. Brox, T. Pock
An iterated L1 Algorithm for Non-smooth Non-convex Optimization in Computer Vision,
Conference on Computer Vision and Pattern Recognition (CVPR), 2013.


Dense Label Interpolation

Download Executable for 64-bit Linux

P. Ochs, T. Brox
Object Segmentation in Video: A Hierarchical Variational Approach for Turning Point Trajectories into Dense Regions,
IEEE International Conference on Computer Vision (ICCV), 2011.


Motion Segmentation

Download Executable for 64-bit Linux (improved pairwise model + densify, PAMI 2013)
Download Code for 64-bit Linux (optical flow variation as used in the definition of the pairwise affinities, PAMI 2013)
Download Executable for 64-bit Linux (higher order, CVPR 2012)
Download Executable for 64-bit Linux (pairwise model, ECCV 2010)

These downloads provide executables with one example video. See the Freiburg Berkeley Motion Segmentation Dataset for the complete dataset.

P. Ochs, J. Malik, T. Brox
Segmentation of moving objects by long term video analysis,
IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(6):1187-1200, Jun. 2014.

P. Ochs, T. Brox
Higher order motion models and spectral clustering,
International Conference on Computer Vision and Pattern Recognition (CVPR), 2012.

T. Brox, J. Malik
Object segmentation by long term analysis of point trajectories,
European Conference on Computer Vision (ECCV), Springer, LNCS, Sept. 2010.


Dense Point Tracking

Download Code with optical flow library for 64-bit Linux
Download Code with optical flow library for Nvidia GPUs (requires CUDA 7.5)
N. Sundaram, T. Brox, K. Keutzer
Dense point trajectories by GPU-accelerated large displacement optical flow,
European Conference on Computer Vision (ECCV), Crete, Greece, Springer, LNCS, Sept. 2010.


Large Displacement Optical Flow

Download Executable for 64-bit Linux
Download C++ Library for 64-bit Linux
Download Executable for 64-bit Mac-OS
Download C++ Library for 64-bit Mac-OSX 10.9 (problems with OSX 10.10)
Download Matlab Mex-functions for 64-bit Linux, 32-bit and 64-bit Windows, and 64-bit Mac-OS

T. Brox, J. Malik
Large displacement optical flow: descriptor matching in variational motion estimation,
IEEE Transactions on Pattern Analysis and Machine Intelligence, 33(3):500-513, March 2011.


Classical Variational Optical Flow

Download Executable for 64-bit Linux
Download C++ Library for 64-bit Linux
Download Executable for 32-bit Windows
Download Matlab Mex-functions for 64-bit Linux, 32-bit and 64-bit Windows

The code is not exactly identical to the work described in the original ECCV 2004 paper. The Windows executable is less efficient and uses an outdated output file format. If you have access to a Linux machine or Matlab, I recommend using these versions.

T. Brox, A. Bruhn, N. Papenberg, J. Weickert
High accuracy optical flow estimation based on a theory for warping,
T. Pajdla and J. Matas (Eds.), European Conference on Computer Vision (ECCV) Prague, Czech Republic, Springer, LNCS, Vol. 3024,  25-36, May 2004.
©Springer-Verlag Berlin Heidelberg 2004
(bibtex)

Nonlocal means with cluster trees

Download Executables for 64-bit Linux

The program runs the non-iterative method described in the paper using no overlap for the cluster tree.

T. Brox, O. Kleinschmidt, D. Cremers
Efficient nonlocal means for denoising of textural patterns,
IEEE Transactions on Image Processing 17(7):1083-1092, July 2008.