Skip to content

janinethoma/soft_contrastive_learning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Soft Contrastive Learning for Visual Localization

This repository contains the code for our NeurIPS 2020 publication Soft Contrastive Learning for Visual Localization.

The corresponding models, training/testing image lists and a movie with visual results can be downloaded here.

This code was tested using TensorFlow 1.10.0 and Python 3.5.6.

It uses the following git repositories as dependencies:

The training data can be downloaded using:

Models used in the paper

Name Model
Off-the-shelf offtheshelf
Triplet trained on Pittsburgh pittsnetvlad
Triplet triplet_xy_000
Quadruplet quadruplet_xy_000
Lazy triplet ha0_lolazy_triplet_muTrue_renone_vl64_pca_eccv_002
Lazy quadruplet ha0_lolazy_quadruplet_muTrue_renone_vl64_pca_eccv_002
Trip.~+ Huber dist. huber_distance_triplet_xy_000
Log-ratio ha0_lologratio_ma15_mi15_muTrue_renone_tu1_vl64_pca_eccv_002
Multi-similarity ha0_loms_loss_msTrue_muTrue_renone_tu1_vl64_pca_eccv_001
Ours al0.8_be15_ha0_lowms_ma15_mi15_msTrue_muTrue_renone_tu1_vl64_pca_eccv_000

About

This repository contains the code for our NeurIPS 2020 publication "Soft Contrastive Learning for Visual Localization".

Topics

Resources

License

Stars

Watchers

Forks

Contributors

Languages