Submission ID: 00639
MKDNet-hardnet
Processed: 20-12-11. Download link: sid-00639-sift8k_mkdnet_hardnet.json
This page ranks the submission against all others using the same number of keypoints, regardless of descriptor size. Please hover over table headers for descriptions on metrics and full scene names.
Metadata
- Authors: (contact)
- Keypoint: sift
- Descriptor: mkdnet-hardnet (128 float32: 512 bytes)
- Number of features: 8000
- Summary: based on the paper [Explicit spatial encoding for deep local descriptors], trained on Liberty set from PhotoTourism dataset
- Paper: https://openaccess.thecvf.com/content_CVPR_2019/papers/Mukundan_Explicit_Spatial_Encoding_for_Deep_Local_Descriptors_CVPR_2019_paper.pdf
- Website: https://openaccess.thecvf.com/content_CVPR_2019/papers/Mukundan_Explicit_Spatial_Encoding_for_Deep_Local_Descriptors_CVPR_2019_paper.pdf
- Origin: Submission
- Flags: is_submission
Phototourism / Stereo track
mAA at 10 degrees: 0.51643 (±0.00099 over 3 run(s) / ±0.14901 over 9 scenes)
Rank (per category): 65 (of 147)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 7969.4 | 577.2 | 577.2 | 381.1 | 0.527 Rank: 30/147 |
0.939 Rank: 69/147 |
0.20816 (±0.00241) Rank: 70/147 |
0.32394 (±0.00213) Rank: 69/147 |
| fcs | 7874.2 | 688.6 | 688.6 | 461.0 | 0.443 Rank: 83/147 |
0.942 Rank: 25/147 |
0.61834 (±0.00111) Rank: 52/147 |
0.74013 (±0.00055) Rank: 50/147 |
| lms | 7770.9 | 480.2 | 480.2 | 252.2 | 0.468 Rank: 53/147 |
0.698 Rank: 47/147 |
0.53922 (±0.00315) Rank: 54/147 |
0.65596 (±0.00304) Rank: 52/147 |
| lb | 7953.6 | 432.8 | 432.8 | 250.9 | 0.474 Rank: 86/147 |
0.765 Rank: 55/147 |
0.42228 (±0.00241) Rank: 70/147 |
0.51246 (±0.00220) Rank: 78/147 |
| mc | 7824.0 | 579.3 | 579.3 | 379.7 | 0.524 Rank: 78/147 |
0.952 Rank: 70/147 |
0.37338 (±0.00241) Rank: 70/147 |
0.52417 (±0.00402) Rank: 80/147 |
| mr | 7797.2 | 525.4 | 525.4 | 377.8 | 0.457 Rank: 120/147 |
0.969 Rank: 56/147 |
0.30801 (±0.00093) Rank: 77/147 |
0.41567 (±0.00123) Rank: 79/147 |
| psm | 7809.6 | 311.4 | 311.4 | 178.9 | 0.385 Rank: 83/147 |
0.591 Rank: 80/147 |
0.14800 (±0.00214) Rank: 69/147 |
0.25991 (±0.00108) Rank: 72/147 |
| sf | 7896.6 | 713.9 | 713.9 | 471.9 | 0.480 Rank: 81/147 |
0.894 Rank: 22/147 |
0.48456 (±0.00230) Rank: 77/147 |
0.62430 (±0.00130) Rank: 70/147 |
| spc | 7869.2 | 561.2 | 561.2 | 324.1 | 0.487 Rank: 94/147 |
0.876 Rank: 43/147 |
0.44036 (±0.00179) Rank: 55/147 |
0.59135 (±0.00230) Rank: 56/147 |
| avg | 7862.7 | 541.1 | 541.1 | 342.0 | 0.472 Rank: 91/147 |
0.847 Rank: 52/147 |
0.39359 (±0.00086) Rank: 65/147 |
0.51643 (±0.00099) Rank: 65/147 |
We show the inliers that survive the robust estimation loop (i.e. RANSAC), or those supplied with the submission if using custom matches, and use the depth estimates to determine whether they are correct. We draw matches above a 5-pixel error threshold in red, and those below are color-coded by their error, from 0 (green) to 5 pixels (yellow). Matches for which we do not have depth estimates are drawn in blue. Please note that the depth maps are estimates and may contain errors.
— british museum —
— florence cathedral side —
— lincoln memorial statue —
— london bridge —
— milan cathedral —
— mount rushmore —
— piazza san marco —
— sagrada familia —
— saint paul's cathedral —
Phototourism / Multiview track
mAA at 10 degrees: 0.69845 (±0.00564 over 3 run(s) / ±0.13555 over 9 scenes)
Rank (per category): 64 (of 147)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 7969.4 | 570.31 | 99.70 Rank: 87/147 |
3633.45 Rank: 25/147 |
4.824 Rank: 108/147 |
0.51312 Rank: 77/147 |
0.43057 (±0.00325) Rank: 79/147 |
0.58120 (±0.00475) Rank: 78/147 |
| fcs | 7874.2 | 721.94 | 98.79 Rank: 23/147 |
5956.49 Rank: 26/147 |
4.420 Rank: 72/147 |
0.25348 Rank: 66/147 |
0.73421 (±0.00159) Rank: 41/147 |
0.79579 (±0.00142) Rank: 56/147 |
| lms | 7770.9 | 522.88 | 99.07 Rank: 70/147 |
4412.80 Rank: 24/147 |
4.354 Rank: 108/147 |
0.27881 Rank: 43/147 |
0.82372 (±0.00290) Rank: 59/147 |
0.87963 (±0.00286) Rank: 60/147 |
| lb | 7953.6 | 529.10 | 98.56 Rank: 40/147 |
3826.44 Rank: 25/147 |
4.492 Rank: 94/147 |
0.50217 Rank: 80/147 |
0.64496 (±0.00485) Rank: 69/147 |
0.74102 (±0.00724) Rank: 69/147 |
| mc | 7824.0 | 568.19 | 99.83 Rank: 39/147 |
4714.14 Rank: 46/147 |
4.505 Rank: 92/147 |
0.40045 Rank: 66/147 |
0.49996 (±0.00583) Rank: 78/147 |
0.64751 (±0.00546) Rank: 79/147 |
| mr | 7797.2 | 530.92 | 94.29 Rank: 67/147 |
4920.68 Rank: 62/147 |
4.100 Rank: 80/147 |
0.48368 Rank: 44/147 |
0.44677 (±0.00926) Rank: 49/147 |
0.56167 (±0.01030) Rank: 47/147 |
| psm | 7809.6 | 303.82 | 92.90 Rank: 61/147 |
4187.48 Rank: 37/147 |
3.069 Rank: 106/147 |
0.60060 Rank: 33/147 |
0.36183 (±0.02051) Rank: 87/147 |
0.45370 (±0.02478) Rank: 81/147 |
| sf | 7896.6 | 706.52 | 98.79 Rank: 34/147 |
6479.27 Rank: 54/147 |
4.378 Rank: 76/147 |
0.31704 Rank: 43/147 |
0.73635 (±0.00529) Rank: 36/147 |
0.82155 (±0.00595) Rank: 33/147 |
| spc | 7869.2 | 576.85 | 99.71 Rank: 31/147 |
5172.30 Rank: 23/147 |
4.270 Rank: 78/147 |
0.43042 Rank: 17/147 |
0.70561 (±0.00390) Rank: 44/147 |
0.80401 (±0.00302) Rank: 42/147 |
| avg | 7862.7 | 558.95 | 97.96 Rank: 44/147 |
4811.45 Rank: 28/147 |
4.268 Rank: 92/147 |
0.41997 Rank: 51/147 |
0.59822 (±0.00491) Rank: 64/147 |
0.69845 (±0.00564) Rank: 64/147 |
In the multi-view track we reconstruct the scene with Structure-from-Motion (Colmap) with small sets of images. We show the results for one bag of 25 images (displaying: 10). Keypoints are drawn in blue if they are part of the model, and in red otherwise.
— british museum —
— florence cathedral side —
— lincoln memorial statue —
— london bridge —
— milan cathedral —
— mount rushmore —
— piazza san marco —
— sagrada familia —
— saint paul's cathedral —
