Submission details
| Method name | DenseVLAD & D2-Net |
| Dataset | robotcar |
| Submission date | Sept. 17, 2019, 1:48 p.m. |
| Last modified | Oct. 31, 2019, 12:58 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 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 | night all | night rain | overcast winter | sun | rain | snow | dawn | dusk | night | day all | overcast summer |
|---|---|---|---|---|---|---|---|---|---|---|---|
| DenseVLAD & D2-Net | 20.4 / 40.1 / 55.0 | 22.7 / 40.5 / 56.1 | 51.8 / 78.5 / 96.2 | 52.0 / 72.4 / 91.1 | 63.7 / 84.3 / 96.9 | 60.5 / 84.5 / 94.9 | 54.0 / 80.3 / 94.4 | 60.2 / 83.0 / 95.9 | 18.0 / 39.7 / 53.9 | 54.5 / 80.0 / 95.3 | 40.4 / 77.3 / 98.3 |