Permalink
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
18 lines (15 sloc) 1.18 KB

Are Large-Scale 3D Models Really Necessary for Accurate Visual Localization?

Torsten Sattler, Akihiko Torii, Josef Sivic, Marc Pollefeys, Hajime Taira, Masatoshi Okutomi, Tomas Pajdla, CVPR,2017

Summary

  • Pipeline:
    • Generating Reference Poses:
      • Generating local reconstructions: generate SfM reconstructions from the database images around the query images.
      • Geo-registration: to obtain the global positions and orientations of the cameras in each local reconstruction.
    • 2D Image-based Localization:
      • Pose Estimation for 2D-based Approaches: ①.Nearest neighbor (NN)/most relevant database image. ②. Spatial re-ranking (SR). ③.SfM on the fly (SfM): small-scale SfM to obtain a local 3D model around the query image.

Strengths / Novelties

  • generate reference camera pose annotations for the query images of the San Francisco Landmarks dataset.
  • use this new dataset for the first comparison of 2Dand 3D-based localization approaches regarding their pose accuracy.
  • Quantitative evaluation of error.