Submission ID: 00041
SuperPoint (2k features, NMS=3), DEGENSAC
Processed: 20-04-23. Download link: sid-00041-superpoint-2k-nms3.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-nms3-r1200
- Descriptor: superpoint-nms3-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 3 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: 0.28679 (±0.00056 over 3 run(s) / ±0.10300 over 9 scenes)
Rank (per category): 88 (of 108)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 2048.0 | 275.7 | 275.7 | 125.1 | 0.417 Rank: 83/108 |
0.816 Rank: 86/108 |
0.14013 (±0.00026) Rank: 57/108 |
0.22508 (±0.00083) Rank: 66/108 |
| fcs | 2048.0 | 283.8 | 283.8 | 130.6 | 0.376 Rank: 27/108 |
0.674 Rank: 94/108 |
0.30541 (±0.00169) Rank: 88/108 |
0.42524 (±0.00189) Rank: 88/108 |
| lms | 2048.0 | 263.8 | 263.8 | 127.6 | 0.381 Rank: 35/108 |
0.555 Rank: 92/108 |
0.32983 (±0.00135) Rank: 86/108 |
0.43483 (±0.00206) Rank: 85/108 |
| lb | 2048.0 | 284.8 | 284.8 | 117.7 | 0.351 Rank: 30/108 |
0.623 Rank: 84/108 |
0.25421 (±0.00026) Rank: 57/108 |
0.35039 (±0.00183) Rank: 73/108 |
| mc | 2048.0 | 270.7 | 270.7 | 108.9 | 0.380 Rank: 33/108 |
0.708 Rank: 92/108 |
0.12932 (±0.00026) Rank: 57/108 |
0.21502 (±0.00286) Rank: 88/108 |
| mr | 2048.0 | 256.0 | 256.0 | 114.5 | 0.418 Rank: 25/108 |
0.719 Rank: 95/108 |
0.11028 (±0.00086) Rank: 88/108 |
0.16709 (±0.00103) Rank: 88/108 |
| psm | 2048.0 | 176.5 | 176.5 | 61.8 | 0.295 Rank: 33/108 |
0.286 Rank: 92/108 |
0.06749 (±0.00079) Rank: 86/108 |
0.14151 (±0.00043) Rank: 85/108 |
| sf | 2048.0 | 345.9 | 345.9 | 171.5 | 0.353 Rank: 30/108 |
0.651 Rank: 89/108 |
0.25040 (±0.00223) Rank: 85/108 |
0.35988 (±0.00182) Rank: 86/108 |
| spc | 2048.0 | 260.2 | 260.2 | 96.5 | 0.342 Rank: 37/108 |
0.615 Rank: 92/108 |
0.16421 (±0.00195) Rank: 88/108 |
0.26204 (±0.00063) Rank: 88/108 |
| avg | 2048.0 | 268.6 | 268.6 | 117.1 | 0.368 Rank: 31/108 |
0.628 Rank: 93/108 |
0.19459 (±0.00085) Rank: 88/108 |
0.28679 (±0.00056) Rank: 88/108 |
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: 0.53829 (±0.00021 over 3 run(s) / ±0.16324 over 9 scenes)
Rank (per category): 82 (of 108)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 2048.0 | 160.61 | 98.44 Rank: 91/108 |
893.42 Rank: 97/108 |
4.588 Rank: 77/108 |
0.67680 Rank: 69/108 |
0.30191 (±0.00885) Rank: 60/108 |
0.44013 (±0.00801) Rank: 63/108 |
| fcs | 2048.0 | 176.39 | 93.76 Rank: 97/108 |
1426.59 Rank: 90/108 |
4.192 Rank: 64/108 |
0.36119 Rank: 92/108 |
0.56161 (±0.00438) Rank: 91/108 |
0.64664 (±0.00779) Rank: 90/108 |
| lms | 2048.0 | 151.86 | 94.54 Rank: 96/108 |
1088.31 Rank: 75/108 |
4.774 Rank: 56/108 |
0.32460 Rank: 63/108 |
0.68215 (±0.00413) Rank: 88/108 |
0.75582 (±0.00665) Rank: 88/108 |
| lb | 2048.0 | 193.55 | 97.14 Rank: 61/108 |
1039.70 Rank: 87/108 |
5.072 Rank: 17/108 |
0.53900 Rank: 76/108 |
0.53933 (±0.00564) Rank: 47/108 |
0.66015 (±0.00792) Rank: 52/108 |
| mc | 2048.0 | 143.25 | 94.93 Rank: 79/108 |
1066.42 Rank: 77/108 |
4.344 Rank: 47/108 |
0.59040 Rank: 92/108 |
0.33331 (±0.01229) Rank: 84/108 |
0.46553 (±0.01652) Rank: 82/108 |
| mr | 2048.0 | 143.33 | 86.56 Rank: 90/108 |
950.05 Rank: 99/108 |
4.256 Rank: 40/108 |
0.68650 Rank: 91/108 |
0.24194 (±0.00574) Rank: 89/108 |
0.33766 (±0.00766) Rank: 89/108 |
| psm | 2048.0 | 73.38 | 74.80 Rank: 75/108 |
763.45 Rank: 99/108 |
2.957 Rank: 61/108 |
0.82884 Rank: 80/108 |
0.18268 (±0.00942) Rank: 72/108 |
0.24964 (±0.00864) Rank: 71/108 |
| sf | 2048.0 | 219.90 | 92.64 Rank: 91/108 |
1729.87 Rank: 71/108 |
4.454 Rank: 50/108 |
0.35442 Rank: 63/108 |
0.61084 (±0.00150) Rank: 80/108 |
0.69379 (±0.00318) Rank: 84/108 |
| spc | 2048.0 | 142.11 | 95.73 Rank: 90/108 |
1132.99 Rank: 96/108 |
4.152 Rank: 76/108 |
0.73641 Rank: 93/108 |
0.47617 (±0.00523) Rank: 93/108 |
0.59522 (±0.00277) Rank: 92/108 |
| avg | 2048.0 | 156.04 | 92.06 Rank: 88/108 |
1121.20 Rank: 96/108 |
4.310 Rank: 54/108 |
0.56646 Rank: 90/108 |
0.43666 (±0.00029) Rank: 83/108 |
0.53829 (±0.00021) Rank: 82/108 |
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 —
