UnrealStereo

UnrealStereo is a project for generating images from virtual worlds for stereo vision. It is based on UnrealCV which is an open source software connecting Computer Vision to Unreal Engine.

This repository contains codes and data for reproducing the results in our paper.

Getting Started

Prerequisites

• Ubuntu

• Python 3

  pip install -r requirements.txt

• Csh (To run the ELAS algorithm)

  sudo apt-get install csh

Installation

1. Clone the repository.

   git clone https://github.com/edz-o/minimum_evaluation.git

   We'll call the directory that you cloned UnrealStereo into $ROOT

Config ELAS algorithm

1. We are using the implementation from Middlebury website.

   wget http://vision.middlebury.edu/stereo/submit3/zip/MiddEval3-SDK-1.6.zip
   unzip MiddEval3-SDK-1.6.zip
   cd MiddEval3

2. Compile Libelas as follows:

   cd alg-ELAS/build
   cmake ..
   make
   cd ../..

3. Compile the tools in code/ as follows:

   cd code/imageLib
   make
   cd ..
   make
   cd ..

4. Copy our python wrapper for the algorithm to MiddEval3/alg-ELAS.

   cd $ROOT
   cp run_ELAS.py MiddEval3/alg-ELAS

Download unrealstereo data

Download the data we used as follows,

wget https://cs.jhu.edu/~yzh/unrealstereo_data_hazardous.zip 
unzip unrealstereo_data_hazardous.zip

The 10k synthetic sequences can be downloaded from here.

Reproduce the result

sh run_all.sh

Annotation of hazardous regions on Middlebury and KITTI 2015

The annotations used in our paper can be downloaded here.

License and Citation

This project is licensed under the MIT License - see the LICENSE file for details.

If you find UnrealStereo useful in your research, please consider citing:

@inproceedings{zhang2018unrealstereo,
  title={Unrealstereo: Controlling hazardous factors to analyze stereo vision},
  author={Zhang, Yi and Qiu, Weichao and Chen, Qi and Hu, Xiaolin and Yuille, Alan},
  booktitle={2018 International Conference on 3D Vision (3DV)},
  pages={228--237},
  year={2018},
  organization={IEEE}
}