Skip to content

prime-slam/evolin

Repository files navigation


tests License: Apache License 2.0

EVOLIN is a benchmark for evaluation of line detection and association results. We provide a set of docker-packed line detection and association algorithms, metrics to evaluate them, and line-annotated data. Additional information can be found on our web page, in the article, and in the documentation.

Installation

Dependencies

  1. Install dependencies
sudo apt update \
&& sudo apt upgrade \
&& sudo apt install --no-install-recommends -y libeigen3-dev cmake
  1. Install our custom g2opy:
git clone https://github.com/anastasiia-kornilova/g2opy
cd g2opy
git checkout lines_opt
mkdir build
cd build
cmake ..
make -j8
cd ..
python setup.py install
  1. Clone this repository
git clone https://github.com/prime-slam/evolin

Annotated data

To evaluate line detectors and associators, we annotated lr kt2 and of kt2 trajectories from ICL NUIM, as well as fr3/cabinet and fr1/desk trajectories from TUM RGB-D. Only breaking segments have been annotated, such as ceilings, floors, walls, doors, and furniture linear elements. The datasets can be downloaded here.

Metrics

The following detection metrics are implemented:

  • Heatmap-based and vectorized classification
    • precision
    • recall
    • F-score
    • average precision
  • Repeatability
    • repeatability score
    • localization error

The following association metrics are implemented:

  • Matching classification
    • precision
    • recall
    • F-score
  • Pose error
    • angular translation error
    • absolute translation error
    • angular rotation error
    • pose error AUC

Get detection and association results

A list of algorithms and instructions for running them can be found in our repository.

Evaluation

The scripts required for evaluation and examples, as well as the documentation are located in evaluation folder. The results of the evaluation of adapted detection and association algorithms can be found in our article.

Cite us

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

@article{evolin2023,
title={EVOLIN Benchmark: Evaluation of Line Detection and Association},
author={Kirill Ivanov, Gonzalo Ferrer, and Anastasiia Kornilova},
journal={arXiv preprint arXiv:2303.05162},
year={2023}}