search

naver / r2d2

Other
467 stars 86 forks source link

Reproduce the results in the paper #4

Closed pengsida closed 4 years ago

pengsida commented 4 years ago

Hi,

Thanks for your excellent paper and code for the feature detection. In this repo, you only released the MMA metric. I want to test the pretrained model on repeatibility, matching score and visual localization. Could you tell me the specific detail on the repeatibility and matching score metric? And could you tell me the localization pipeline used in your experiment?

humenbergerm commented 4 years ago

Hi!

For visual localization, we used this pipeline: https://github.com/tsattler/visuallocalizationbenchmark/tree/master/local_feature_evaluation It was part of the local feature task of the LTVL challenge at CVPR19.

Best, Martin

pengsida commented 4 years ago

Thank you!