Submission details
| Method name | DenseVLAD & D2-Net |
| Dataset | extended-cmu |
| Submission date | Sept. 20, 2019, 5:28 p.m. |
| Last modified | Feb. 25, 2020, 9:10 p.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 DenseVLAD (single-scale) to identify the top-20 closest database images, then uses single-scale D2-Net features for feature matching and COLMAP's image_registrator for localization. Only "rear" images are considered. |
Detailed results for all conditions
| Method | urban | suburban | park | overcast | sunny | foliage | mixed foliage | no foliage | low sun | cloudy | snow |
|---|---|---|---|---|---|---|---|---|---|---|---|
| DenseVLAD & D2-Net | 94.0 / 97.7 / 99.1 | 93.0 / 95.7 / 98.3 | 89.2 / 93.2 / 95.0 | 92.5 / 95.6 / 97.5 | 86.2 / 91.8 / 95.2 | 88.0 / 92.8 / 95.9 | 94.3 / 96.9 / 98.4 | 98.0 / 99.4 / 99.8 | 95.1 / 97.5 / 98.7 | 95.5 / 97.5 / 98.9 | 97.2 / 98.9 / 99.6 |