Submission ID: ciffa4wz
Example: Upright SIFT (OpenCV), 8k features
Processed: 2021-05-10. Download link: ciffa4wz6qbzkp3czhfxpc-rootsift-upright-8k-both-degensac.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.
- Phototourism dataset: Stereo track / Multiview track
- Prague Parks dataset: Stereo track / Multiview track
- Google Urban dataset: Stereo track / Multiview track
Metadata
- Authors: (contact)
- Keypoint: sift-def
- Descriptor: rootsift-upright (128 float32: 512 bytes)
- Number of features: 8000
- Summary: SIFT with 8000 features, using the built-in matcher (bidirectional filter with the both strategy, optimal inlier and ratio test thresholds) with DEGENSAC, and setting keypoint orientation to a constant value to increase performance.
- Paper: N/A
- Website: https://opencv.org
- Processing date: 2021-05-10
Phototourism dataset / Stereo track
mAA at 10 degrees: 0.46173 (±0.00000 over 1 run(s) / ±0.12984 over 9 scenes)
Rank (per category): 38 (of 51)
Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
BM | 3130.8 | 258.6 | 180.4 | 0.450 Rank: 50/51 |
0.962 Rank: 2/51 |
0.21197 (±0.00000) Rank: 31/51 |
0.33652 (±0.00000) Rank: 31/51 |
FCS | 6447.4 | 469.6 | 337.9 | 0.418 Rank: 50/51 |
0.901 Rank: 18/51 |
0.57777 (±0.00000) Rank: 31/51 |
0.69310 (±0.00000) Rank: 32/51 |
LMS | 1067.8 | 107.4 | 71.7 | 0.360 Rank: 44/51 |
0.700 Rank: 14/51 |
0.38350 (±0.00000) Rank: 40/51 |
0.50275 (±0.00000) Rank: 40/51 |
LB | 3028.0 | 196.0 | 122.0 | 0.367 Rank: 50/51 |
0.721 Rank: 27/51 |
0.35736 (±0.00000) Rank: 31/51 |
0.44272 (±0.00000) Rank: 38/51 |
MC | 4610.4 | 360.7 | 260.2 | 0.446 Rank: 50/51 |
0.968 Rank: 3/51 |
0.37166 (±0.00000) Rank: 31/51 |
0.51889 (±0.00000) Rank: 33/51 |
MR | 4695.2 | 443.8 | 341.5 | 0.405 Rank: 51/51 |
0.975 Rank: 4/51 |
0.31171 (±0.00000) Rank: 26/51 |
0.42119 (±0.00000) Rank: 25/51 |
PSM | 4834.1 | 168.8 | 105.2 | 0.338 Rank: 50/51 |
0.619 Rank: 37/51 |
0.11405 (±0.00000) Rank: 37/51 |
0.20299 (±0.00000) Rank: 38/51 |
SF | 5954.8 | 591.4 | 422.2 | 0.429 Rank: 41/51 |
0.827 Rank: 26/51 |
0.43007 (±0.00000) Rank: 36/51 |
0.55132 (±0.00000) Rank: 36/51 |
SPC | 4349.3 | 246.9 | 152.3 | 0.400 Rank: 50/51 |
0.822 Rank: 28/51 |
0.35811 (±0.00000) Rank: 40/51 |
0.48608 (±0.00000) Rank: 40/51 |
Avg | 4235.3 | 315.9 | 221.5 | 0.401 Rank: 50/51 |
0.833 Rank: 22/51 |
0.34624 (±0.00000) Rank: 38/51 |
0.46173 (±0.00000) Rank: 38/51 |
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 dataset / Multiview track
mAA at 10 degrees: N/A (±0.00000 over 1 run(s) / ±0.14363 over 9 scenes)
Rank (per category): 39 (of 51)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
BM | 3130.8 | 260.09 | 99.08 Rank: 45/51 |
1448.04 Rank: 43/51 |
4.848 Rank: 31/51 |
0.62135 Rank: 45/51 |
0.35088 (±0.00000) Rank: 45/51 |
0.49050 (±0.00000) Rank: 45/51 |
FCS | 6447.4 | 494.02 | 96.73 Rank: 32/51 |
4216.28 Rank: 32/51 |
4.295 Rank: 34/51 |
0.26440 Rank: 25/51 |
0.69512 (±0.00000) Rank: 31/51 |
0.76649 (±0.00000) Rank: 32/51 |
LMS | 1067.8 | 117.75 | 96.70 Rank: 43/51 |
826.10 Rank: 44/51 |
4.513 Rank: 29/51 |
0.31656 Rank: 26/51 |
0.72351 (±0.00000) Rank: 43/51 |
0.80789 (±0.00000) Rank: 39/51 |
LB | 3028.0 | 242.01 | 96.66 Rank: 42/51 |
1705.85 Rank: 42/51 |
4.322 Rank: 38/51 |
0.53485 Rank: 43/51 |
0.53118 (±0.00000) Rank: 44/51 |
0.63574 (±0.00000) Rank: 45/51 |
MC | 4610.4 | 349.17 | 99.87 Rank: 23/51 |
2773.57 Rank: 41/51 |
4.410 Rank: 40/51 |
0.40835 Rank: 39/51 |
0.50051 (±0.00000) Rank: 39/51 |
0.64674 (±0.00000) Rank: 39/51 |
MR | 4695.2 | 444.48 | 94.00 Rank: 33/51 |
3652.77 Rank: 36/51 |
4.127 Rank: 37/51 |
0.57839 Rank: 45/51 |
0.39301 (±0.00000) Rank: 37/51 |
0.50586 (±0.00000) Rank: 37/51 |
PSM | 4834.1 | 163.83 | 86.63 Rank: 37/51 |
2067.59 Rank: 48/51 |
3.105 Rank: 40/51 |
0.64925 Rank: 39/51 |
0.27670 (±0.00000) Rank: 40/51 |
0.35244 (±0.00000) Rank: 40/51 |
SF | 5954.8 | 584.63 | 95.67 Rank: 37/51 |
5315.32 Rank: 30/51 |
4.259 Rank: 39/51 |
0.32605 Rank: 31/51 |
0.67013 (±0.00000) Rank: 36/51 |
0.75696 (±0.00000) Rank: 35/51 |
SPC | 4349.3 | 252.80 | 95.79 Rank: 41/51 |
2320.63 Rank: 43/51 |
4.126 Rank: 38/51 |
0.50671 Rank: 31/51 |
0.61549 (±0.00000) Rank: 40/51 |
0.71395 (±0.00000) Rank: 39/51 |
Avg | 4235.3 | 323.20 | 95.68 Rank: 37/51 |
2702.91 Rank: 41/51 |
4.223 Rank: 38/51 |
0.46732 Rank: 40/51 |
0.52850 (±0.00000) Rank: 43/51 |
0.63073 (±0.00000) Rank: 39/51 |
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 10 images. 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 —










Prague Parks dataset / Stereo track
mAA at 10 degrees: 0.53140 (±0.00000 over 1 run(s) / ±0.04655 over 3 scenes)
Rank (per category): 31 (of 51)
Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
Lizard | 7995.3 | 573.1 | 456.4 | 0.145 Rank: 1/51 |
0.010 Rank: 21/51 |
0.36560 (±0.00000) Rank: 28/51 |
0.47406 (±0.00000) Rank: 29/51 |
Pond | 7961.0 | 750.3 | 541.2 | 0.164 Rank: 13/51 |
0.055 Rank: 13/51 |
0.42044 (±0.00000) Rank: 29/51 |
0.53206 (±0.00000) Rank: 29/51 |
Tree | 7888.8 | 485.1 | 318.7 | 0.129 Rank: 16/51 |
0.019 Rank: 28/51 |
0.50464 (±0.00000) Rank: 28/51 |
0.58808 (±0.00000) Rank: 30/51 |
Avg | 7948.4 | 602.8 | 438.8 | 0.146 Rank: 13/51 |
0.028 Rank: 13/51 |
0.43023 (±0.00000) Rank: 29/51 |
0.53140 (±0.00000) Rank: 31/51 |
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.
— Lizard —










— Pond —










— Tree —










Prague Parks dataset / Multiview track
mAA at 10 degrees: N/A (±0.00000 over 1 run(s) / ±0.14399 over 3 scenes)
Rank (per category): 15 (of 51)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
Lizard | 7995.3 | 454.48 | 81.37 Rank: 35/51 |
1324.83 Rank: 30/51 |
3.196 Rank: 28/51 |
13.21633 Rank: 4/51 |
0.56200 (±0.00000) Rank: 17/51 |
0.62680 (±0.00000) Rank: 21/51 |
Pond | 7961.0 | 800.40 | 66.80 Rank: 35/51 |
1600.82 Rank: 29/51 |
2.798 Rank: 30/51 |
0.40621 Rank: 5/51 |
0.24693 (±0.00000) Rank: 31/51 |
0.28264 (±0.00000) Rank: 29/51 |
Tree | 7888.8 | 448.03 | 71.67 Rank: 17/51 |
959.94 Rank: 30/51 |
2.944 Rank: 20/51 |
5.76963 Rank: 10/51 |
0.47622 (±0.00000) Rank: 7/51 |
0.52152 (±0.00000) Rank: 3/51 |
Avg | 7948.4 | 567.64 | 73.28 Rank: 34/51 |
1295.20 Rank: 30/51 |
2.980 Rank: 28/51 |
6.46406 Rank: 4/51 |
0.42839 (±0.00000) Rank: 13/51 |
0.47699 (±0.00000) Rank: 15/51 |
Google Urban dataset / Stereo track
mAA at 10 degrees: 0.25922 (±0.00000 over 1 run(s) / ±0.12157 over 17 scenes)
Rank (per category): 25 (of 51)
Scene | Features | Matches (raw) |
Matches (final) |
mAA(5o) | mAA(10o) |
Amsterdam (AMS) | 1994.1 | 167.6 | 104.2 | 0.12857 (±0.00000) Rank: 26/51 |
0.24330 (±0.00000) Rank: 26/51 |
Bangkok (BGK) | 1603.2 | 145.3 | 85.0 | 0.04119 (±0.00000) Rank: 24/51 |
0.12184 (±0.00000) Rank: 25/51 |
Barcelona (BCN) | 1821.4 | 130.1 | 81.0 | 0.09304 (±0.00000) Rank: 22/51 |
0.16154 (±0.00000) Rank: 23/51 |
Buenos Aires (BAR) | 2040.9 | 170.6 | 98.8 | 0.12694 (±0.00000) Rank: 26/51 |
0.21644 (±0.00000) Rank: 26/51 |
Cambridge (CAM) | 1290.4 | 100.2 | 53.3 | 0.22123 (±0.00000) Rank: 26/51 |
0.33904 (±0.00000) Rank: 19/51 |
Cannes (CAN) | 1219.3 | 129.4 | 77.0 | 0.31521 (±0.00000) Rank: 22/51 |
0.40046 (±0.00000) Rank: 23/51 |
Chicago (CHI) | 1453.1 | 62.2 | 24.6 | 0.08115 (±0.00000) Rank: 19/51 |
0.13682 (±0.00000) Rank: 19/51 |
Helsinki (HEL) | 1781.5 | 207.6 | 127.2 | 0.31392 (±0.00000) Rank: 23/51 |
0.51289 (±0.00000) Rank: 22/51 |
Madrid (MAD) | 2017.7 | 120.1 | 55.2 | 0.04498 (±0.00000) Rank: 24/51 |
0.10182 (±0.00000) Rank: 24/51 |
Mountain View (MTV) | 1630.9 | 91.0 | 39.5 | 0.11445 (±0.00000) Rank: 23/51 |
0.18867 (±0.00000) Rank: 25/51 |
New Orleans (NOR) | 1694.7 | 115.3 | 58.2 | 0.16686 (±0.00000) Rank: 25/51 |
0.22485 (±0.00000) Rank: 25/51 |
San Francisco (SF) | 1580.9 | 117.7 | 66.3 | 0.06929 (±0.00000) Rank: 25/51 |
0.17192 (±0.00000) Rank: 24/51 |
Singapore (SG) | 1665.1 | 141.2 | 74.3 | 0.22342 (±0.00000) Rank: 19/51 |
0.31076 (±0.00000) Rank: 21/51 |
Sydney (SYD) | 1869.1 | 134.1 | 65.1 | 0.19738 (±0.00000) Rank: 26/51 |
0.30437 (±0.00000) Rank: 25/51 |
Tokyo (TOK) | 1574.9 | 154.4 | 79.6 | 0.37451 (±0.00000) Rank: 23/51 |
0.51495 (±0.00000) Rank: 22/51 |
Toronto (TOR) | 1867.8 | 101.6 | 40.3 | 0.08240 (±0.00000) Rank: 27/51 |
0.18080 (±0.00000) Rank: 26/51 |
Zurich (ZRH) | 1848.8 | 155.5 | 97.8 | 0.21237 (±0.00000) Rank: 22/51 |
0.27629 (±0.00000) Rank: 23/51 |
Average (Avg) | 1703.2 | 132.0 | 72.2 | 0.16511 (±0.00000) Rank: 24/51 |
0.25922 (±0.00000) Rank: 25/51 |
We show the inliers that survive the robust estimation loop (i.e. RANSAC), or those supplied with the submission if using custom matches. As the dataset does not contain depth, we color them based on the symmetric epopilar distance on normalized coordinates: from 0 (green) to 2e-4 (yellow), and in red above it. Note that these occasionally contain false positives.
— Amsterdam —










— Bangkok —










— Barcelona —










— Buenos Aires —










— Cambridge —










— Cannes —










— Chicago —










— Helsinki —










— Madrid —










— Mountain View —










— New Orleans —










— San Francisco —










— Singapore —










— Sydney —










— Tokyo —










— Toronto —










— Zurich —










Google Urban dataset / Multiview track
mAA at 10 degrees: N/A (±0.00000 over 1 run(s) / ±0.08283 over 17 scenes)
Rank (per category): 28 (of 51)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
AMS | 1994.1 | 73.77 | 50.71 Rank: 28/51 |
326.40 Rank: 28/51 |
2.892 Rank: 27/51 |
6.94109 Rank: 1/51 |
0.02253 (±0.00000) Rank: 28/51 |
0.04422 (±0.00000) Rank: 28/51 |
BGK | 1603.2 | 67.89 | 81.42 Rank: 28/51 |
629.74 Rank: 28/51 |
3.306 Rank: 22/51 |
9.08912 Rank: 1/51 |
0.01813 (±0.00000) Rank: 24/51 |
0.05247 (±0.00000) Rank: 25/51 |
BCN | 1821.4 | 57.02 | 59.83 Rank: 28/51 |
333.10 Rank: 28/51 |
3.019 Rank: 26/51 |
17.77140 Rank: 1/51 |
0.03047 (±0.00000) Rank: 28/51 |
0.05350 (±0.00000) Rank: 28/51 |
BAR | 2040.9 | 71.07 | 48.32 Rank: 28/51 |
233.67 Rank: 28/51 |
2.624 Rank: 27/51 |
4.94063 Rank: 1/51 |
0.00057 (±0.00000) Rank: 28/51 |
0.00311 (±0.00000) Rank: 28/51 |
CAM | 1290.4 | 41.32 | 44.12 Rank: 28/51 |
153.05 Rank: 28/51 |
2.713 Rank: 26/51 |
5.84526 Rank: 1/51 |
0.00794 (±0.00000) Rank: 28/51 |
0.01496 (±0.00000) Rank: 28/51 |
CAN | 1219.3 | 56.27 | 45.77 Rank: 28/51 |
163.83 Rank: 28/51 |
2.858 Rank: 22/51 |
5.61758 Rank: 1/51 |
0.00675 (±0.00000) Rank: 27/51 |
0.01851 (±0.00000) Rank: 27/51 |
CHI | 1453.1 | 30.30 | 39.11 Rank: 28/51 |
96.68 Rank: 28/51 |
2.697 Rank: 28/51 |
23.11980 Rank: 1/51 |
0.00113 (±0.00000) Rank: 28/51 |
0.00371 (±0.00000) Rank: 28/51 |
HEL | 1781.5 | 113.78 | 78.95 Rank: 28/51 |
908.18 Rank: 28/51 |
3.094 Rank: 25/51 |
9.14733 Rank: 1/51 |
0.18536 (±0.00000) Rank: 28/51 |
0.29516 (±0.00000) Rank: 28/51 |
MAD | 2017.7 | 50.40 | 59.92 Rank: 28/51 |
331.68 Rank: 28/51 |
2.873 Rank: 27/51 |
4.27527 Rank: 8/51 |
0.00788 (±0.00000) Rank: 28/51 |
0.02240 (±0.00000) Rank: 28/51 |
MTV | 1630.9 | 43.08 | 39.49 Rank: 28/51 |
182.04 Rank: 28/51 |
2.502 Rank: 28/51 |
2.61535 Rank: 1/51 |
0.00199 (±0.00000) Rank: 28/51 |
0.00503 (±0.00000) Rank: 28/51 |
NOR | 1694.7 | 52.00 | 38.19 Rank: 28/51 |
192.65 Rank: 28/51 |
2.540 Rank: 26/51 |
3.11124 Rank: 2/51 |
0.00252 (±0.00000) Rank: 28/51 |
0.00626 (±0.00000) Rank: 28/51 |
SF | 1580.9 | 50.68 | 38.44 Rank: 28/51 |
93.23 Rank: 28/51 |
2.519 Rank: 28/51 |
1.54766 Rank: 1/51 |
0.00053 (±0.00000) Rank: 28/51 |
0.00272 (±0.00000) Rank: 28/51 |
SG | 1665.1 | 64.72 | 49.89 Rank: 28/51 |
244.48 Rank: 28/51 |
2.905 Rank: 22/51 |
7.65799 Rank: 1/51 |
0.01764 (±0.00000) Rank: 28/51 |
0.03540 (±0.00000) Rank: 28/51 |
SYD | 1869.1 | 58.41 | 54.26 Rank: 28/51 |
230.53 Rank: 28/51 |
2.686 Rank: 27/51 |
5.52509 Rank: 1/51 |
0.02695 (±0.00000) Rank: 28/51 |
0.05126 (±0.00000) Rank: 28/51 |
TOK | 1574.9 | 72.82 | 67.02 Rank: 28/51 |
526.81 Rank: 28/51 |
3.093 Rank: 23/51 |
5.39339 Rank: 1/51 |
0.20620 (±0.00000) Rank: 28/51 |
0.25230 (±0.00000) Rank: 28/51 |
TOR | 1867.8 | 40.21 | 53.90 Rank: 27/51 |
145.47 Rank: 28/51 |
3.187 Rank: 9/51 |
6.68789 Rank: 1/51 |
0.00487 (±0.00000) Rank: 28/51 |
0.01090 (±0.00000) Rank: 28/51 |
ZRH | 1848.8 | 68.97 | 52.52 Rank: 28/51 |
430.05 Rank: 28/51 |
2.768 Rank: 26/51 |
7.47876 Rank: 1/51 |
0.01410 (±0.00000) Rank: 27/51 |
0.02857 (±0.00000) Rank: 28/51 |
Avg | 1703.2 | 59.57 | 53.05 Rank: 28/51 |
307.15 Rank: 28/51 |
2.840 Rank: 27/51 |
7.45676 Rank: 1/51 |
0.03268 (±0.00000) Rank: 28/51 |
0.05297 (±0.00000) Rank: 28/51 |
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 10 images. Keypoints are drawn in blue if they are part of the model, and in red otherwise.
— Amsterdam —










— Bangkok —










— Barcelona —










— Buenos Aires —










— Cambridge —










— Cannes —










— Chicago —










— Helsinki —










— Madrid —










— Mountain View —










— New Orleans —










— San Francisco —










— Singapore —










— Sydney —










— Tokyo —










— Toronto —










— Zurich —









