request the experiment code

mihaidusmanu / d2-net

D2-Net: A Trainable CNN for Joint Description and Detection of Local Features

Other

761 stars 163 forks source link

request the experiment code #63

Closed fantuslk closed 4 years ago

fantuslk commented 4 years ago

Hi mihaidusmanu,could you please give me the code of the experiment in the paper, such as 5.1 image matching, 5.2 3d reconstruction, 5.3localization under chanllenging conditions.I found it difficult to implement the code by myself.Thank you!

mihaidusmanu commented 4 years ago

The code is already available online:

Image matching: https://github.com/mihaidusmanu/d2-net/tree/master/hpatches_sequences
3D reconstruction: https://github.com/ahojnnes/local-feature-evaluation
Aachen Day-Night: https://github.com/tsattler/visuallocalizationbenchmark/tree/master/local_feature_evaluation
InLoc: https://github.com/mihaidusmanu/d2-net/tree/master/inloc

fantuslk commented 4 years ago

Thank you very much.

fantuslk commented 4 years ago

hi，@mihaidusmanu,in the Aachen Daynight's code,only get the result of (0.5m,2),(1m,5),(5m,10),how can I get the result of the features and (10m,25) in the paper? Correctly localized queries (%) Method # Features 0.5m, 2◦ 1.0m, 5◦ 5.0m, 10◦ 10m, 25◦ Upright RootSIFT [29] 11.3K 36.7 54.1 72.5 81.6...

mihaidusmanu commented 4 years ago

Yes, only results for ((0.5m, 2), (1m, 5), (5m, 10)) are currently available to the evaluation system. You might want to contact the authors of the long-term visual localization benchmark for plots or results under different threshold. You can also try to open an issue at https://github.com/tsattler/visuallocalizationbenchmark.