Awesome

{J, T}-Linkage for Vanishing Point Estimation

This is an implementation of J-Linkage [1] and T-Linkage [2] for vanishing point estimation from line segments extracted via LSD [3].

This implementation was used in our CONSAC paper [4], so please cite the paper if you use this code:

@inproceedings{kluger2020consac,
  title={CONSAC: Robust Multi-Model Fitting by Conditional Sample Consensus},
  author={Kluger, Florian and Brachmann, Eric and Ackermann, Hanno and Rother, Carsten and Yang, Michael Ying and Rosenhahn, Bodo},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2020}
}

Setup

Assuming that you are using Anaconda.

Get the code:

git clone --recurse-submodules https://github.com/fkluger/vp-linkage.git
cd vp-linkage
git submodule update --init --recursive

Prepare environment:

conda env create -f environment.yml
source activate vp_linkage
cd datasets/nyu_vp/lsd
python setup.py build_ext --inplace
cd ../../yud_plus/lsd
python setup.py build_ext --inplace
cd ../../..

Datasets

NYU-VP

The vanishing point labels and pre-extracted line segments for the NYU dataset are fetched automatically via the nyu_vp submodule. In order to use the original RGB images as well, you need to obtain the original dataset MAT-file and convert it to a version 7 MAT-file in MATLAB so that we can load it via scipy:

load('nyu_depth_v2_labeled.mat')
save('nyu_depth_v2_labeled.v7.mat','-v7')

YUD and YUD+

Pre-extracted line segments and VP labels are fetched automatically via the yud_plus submodule. RGB images and camera calibration parameters, however, are not included. Download the original York Urban Dataset from the Elder Laboratory's website and store it under the datasets/yud_plus/data subfolder.

Demo

...coming soon(ish)

Run

To compute the AUC metric over the YUD test set, run:

python linkage.py --dataset yud --dataset_path ./datasets/yud_plus/data/ 

For YUD+:

python linkage.py --dataset yud+ --dataset_path ./datasets/yud_plus/data/ 

For NYU-VP:

python linkage.py --dataset nyu --dataset_path ./datasets/nyu_vp/data/ --mat_file_path nyu_depth_v2_labeled.v7.mat

Add the option --tlinkage in order to switch from J-Linkage to T-Linkage. See python linkage.py --help for available options.

References

[1] Roberto Toldo and Andrea Fusiello. Robust multiple structures estimation with j-linkage. ECCV 2008.

[2] Luca Magri and Andrea Fusiello. T-linkage: A continuous relaxation of j-linkage for multi-model fitting. CVPR 2014.

[3] Rafael Grompone Von Gioi, Jeremie Jakubowicz, Jean-Michel Morel, and Gregory Randall. Lsd: A fast line segment detector with a false detection control. TPAMI 2008.

[4] Florian Kluger, Eric Brachmann, Hanno Ackermann, Carsten Rother, Michael Ying Yang, and Bodo Rosenhahn. CONSAC: Robust Multi-Model Fitting by Conditional Sample Consensus. CVPR 2020