Submission details
Method name | NetVLAD (top-50) & D2-Net - single-scale |
Dataset | silda |
Submission date | Sept. 26, 2019, 10:28 a.m. |
Last modified | Sept. 26, 2019, 10:28 a.m. |
Publication url | http://openaccess.thecvf.com/content_CVPR_2019/papers/Dusmanu_D2-Net_A_Trainable_CNN_for_Joint_Description_and_Detection_of_CVPR_2019_paper.pdf |
Code url | https://github.com/mihaidusmanu/d2-net |
Other info | Uses NetVLAD (sum_edges = 1400) to identify the top-50 closest database images, then uses single-scale D2-Net features matching and COLMAP's image_registrator for localization. |
Detailed results for all conditions
Method | evening | snow | night |
---|---|---|---|
NetVLAD (top-50) & D2-Net - single-scale | 30.0 / 67.8 / 94.4 | 1.4 / 10.8 / 67.5 | 25.7 / 51.9 / 79.9 |