Submission ID: 00031
SuperPoint (2k features, NMS=2), DEGENSAC
Processed: 20-04-22. Download link: sid-00031-superpoint-2k-nms2.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: Challenge organizers (contact)
- Keypoint: superpoint-nms2-r1200
- Descriptor: superpoint-nms2-r1200 (256 float32: 1024 bytes)
- Number of features: 2048
- Summary: SuperPoint (initial release based on Pytorch). Lowered detection threshold to obtain up to 2k features per image. Using the built-in Non-Maxima Suppression filter at 2 pixels. Images are resized to 1200 pixels on the largest side.
- Paper: https://arxiv.org/abs/1712.07629
- Website: https://github.com/MagicLeapResearch/SuperPointPretrainedNetwork
- Origin: Baseline
- Flags: is_baseline
Phototourism / Stereo track
mAA at 10 degrees: 0.27954 (±0.00073 over 3 run(s) / ±0.10151 over 9 scenes)
Rank (per category): 90 (of 108)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 2048.0 | 234.6 | 234.6 | 111.3 | 0.427 Rank: 81/108 |
0.819 Rank: 85/108 |
0.13472 (±0.00177) Rank: 59/108 |
0.21862 (±0.00157) Rank: 73/108 |
| fcs | 2048.0 | 261.3 | 261.3 | 122.3 | 0.381 Rank: 25/108 |
0.669 Rank: 95/108 |
0.30326 (±0.00229) Rank: 89/108 |
0.42246 (±0.00252) Rank: 89/108 |
| lms | 2048.0 | 214.1 | 214.1 | 111.1 | 0.396 Rank: 17/108 |
0.555 Rank: 93/108 |
0.31482 (±0.00225) Rank: 89/108 |
0.41642 (±0.00101) Rank: 89/108 |
| lb | 2048.0 | 249.0 | 249.0 | 107.0 | 0.358 Rank: 29/108 |
0.620 Rank: 86/108 |
0.24547 (±0.00177) Rank: 59/108 |
0.34071 (±0.00201) Rank: 77/108 |
| mc | 2048.0 | 244.4 | 244.4 | 100.5 | 0.383 Rank: 31/108 |
0.706 Rank: 93/108 |
0.12456 (±0.00177) Rank: 59/108 |
0.21072 (±0.00114) Rank: 89/108 |
| mr | 2048.0 | 209.9 | 209.9 | 96.1 | 0.436 Rank: 17/108 |
0.703 Rank: 97/108 |
0.10572 (±0.00138) Rank: 89/108 |
0.15950 (±0.00184) Rank: 91/108 |
| psm | 2048.0 | 157.4 | 157.4 | 56.7 | 0.300 Rank: 32/108 |
0.281 Rank: 93/108 |
0.06613 (±0.00127) Rank: 89/108 |
0.13631 (±0.00117) Rank: 88/108 |
| sf | 2048.0 | 310.1 | 310.1 | 156.9 | 0.355 Rank: 27/108 |
0.650 Rank: 91/108 |
0.24492 (±0.00131) Rank: 87/108 |
0.35611 (±0.00099) Rank: 87/108 |
| spc | 2048.0 | 228.6 | 228.6 | 87.8 | 0.348 Rank: 32/108 |
0.615 Rank: 93/108 |
0.15757 (±0.00326) Rank: 90/108 |
0.25500 (±0.00290) Rank: 90/108 |
| avg | 2048.0 | 234.4 | 234.4 | 105.5 | 0.376 Rank: 30/108 |
0.624 Rank: 94/108 |
0.18857 (±0.00111) Rank: 89/108 |
0.27954 (±0.00073) Rank: 90/108 |
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: 0.51061 (±0.00378 over 3 run(s) / ±0.16592 over 9 scenes)
Rank (per category): 90 (of 108)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 2048.0 | 132.32 | 97.81 Rank: 93/108 |
760.74 Rank: 103/108 |
4.584 Rank: 78/108 |
0.69458 Rank: 74/108 |
0.27336 (±0.01075) Rank: 77/108 |
0.40656 (±0.01401) Rank: 79/108 |
| fcs | 2048.0 | 157.02 | 93.46 Rank: 98/108 |
1352.81 Rank: 96/108 |
4.111 Rank: 73/108 |
0.34370 Rank: 90/108 |
0.54611 (±0.00948) Rank: 93/108 |
0.63377 (±0.01174) Rank: 93/108 |
| lms | 2048.0 | 117.02 | 93.02 Rank: 99/108 |
914.72 Rank: 98/108 |
4.668 Rank: 70/108 |
0.37351 Rank: 91/108 |
0.62903 (±0.00915) Rank: 92/108 |
0.70488 (±0.00969) Rank: 93/108 |
| lb | 2048.0 | 160.94 | 96.29 Rank: 74/108 |
946.19 Rank: 93/108 |
4.981 Rank: 21/108 |
0.54702 Rank: 81/108 |
0.52645 (±0.00617) Rank: 54/108 |
0.64542 (±0.00359) Rank: 53/108 |
| mc | 2048.0 | 125.19 | 94.21 Rank: 85/108 |
984.18 Rank: 90/108 |
4.242 Rank: 51/108 |
0.59266 Rank: 93/108 |
0.31271 (±0.00897) Rank: 89/108 |
0.43958 (±0.01237) Rank: 88/108 |
| mr | 2048.0 | 109.15 | 80.78 Rank: 103/108 |
766.00 Rank: 103/108 |
3.910 Rank: 79/108 |
0.87964 Rank: 99/108 |
0.19791 (±0.00317) Rank: 93/108 |
0.28420 (±0.00477) Rank: 93/108 |
| psm | 2048.0 | 62.84 | 72.43 Rank: 87/108 |
680.54 Rank: 105/108 |
2.925 Rank: 70/108 |
0.76604 Rank: 68/108 |
0.16826 (±0.00564) Rank: 74/108 |
0.23124 (±0.00802) Rank: 76/108 |
| sf | 2048.0 | 190.35 | 91.83 Rank: 97/108 |
1587.50 Rank: 88/108 |
4.401 Rank: 63/108 |
0.36769 Rank: 71/108 |
0.59109 (±0.00534) Rank: 86/108 |
0.67667 (±0.00455) Rank: 88/108 |
| spc | 2048.0 | 119.88 | 94.17 Rank: 99/108 |
1025.61 Rank: 103/108 |
4.080 Rank: 81/108 |
0.77442 Rank: 94/108 |
0.46273 (±0.00592) Rank: 94/108 |
0.57318 (±0.00629) Rank: 94/108 |
| avg | 2048.0 | 130.52 | 90.45 Rank: 96/108 |
1002.03 Rank: 102/108 |
4.211 Rank: 71/108 |
0.59325 Rank: 93/108 |
0.41196 (±0.00225) Rank: 89/108 |
0.51061 (±0.00378) Rank: 90/108 |
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 —
