Skip to content

Latest commit

 

History

History
61 lines (39 loc) · 3.1 KB

benchmark_results.adoc

File metadata and controls

61 lines (39 loc) · 3.1 KB

kapture-localization / Image Retrieval Benchmark results

Table of Contents

Introduction

Here we present additional results (more global features, different local features) computed to our image retrieval for visual localization benchmark presented in these two papers: 3DV20, IJCV21. Wrt local features, we use two versions of R2D2: (i) the same as in the papers and (ii) a novel version which is not yet publicly available. However, we provide the local features for all datasets via our dataset downloader.

Furthermore, we used a new implementation of the local_sfm (paradigm 2a in the papers) method. This version is much faster and provides similar performance.

If you use these results in your own work, please cite one of our papers:

@inproceedings{benchmarking_ir3DV2020,
      title={Benchmarking Image Retrieval for Visual Localization},
      author={No{\'e} Pion, Martin Humenberger, Gabriela Csurka, Yohann Cabon, Torsten Sattler},
      year={2020},
      booktitle={International Conference on 3D Vision}
}

@article{humenberger2022investigating,
  title={Investigating the Role of Image Retrieval for Visual Localization},
  author={Humenberger, Martin and Cabon, Yohann and Pion, No{\'e} and Weinzaepfel, Philippe and Lee, Donghwan and Gu{\'e}rin, Nicolas and Sattler, Torsten and Csurka, Gabriela},
  journal={International Journal of Computer Vision},
  year={2022},
  publisher={Springer}
}

Image retrieval models

AP-GeM-LM18: model, code

r101delg_gldv2clean: model

densevlad_multi: paper, code

fire: model, code

netvlad_vd16pitts: code, model

openibl_vgg16_netvlad: code, model