Submission ID: b48d019f
SPG
Processed: 21-06-10. Download link: b48d019f5dc367be-spg.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: OPPOZuiShuaiDeSanGe (contact)
- Keypoint: spspspsp
- Descriptor: spspspsp (256 float32: 1024 bytes)
- Number of features: 2048
- Summary: Sp and Sg result 0531
- Paper: N/A
- Website: N/A
- Processing date: 21-06-10
Phototourism dataset / Stereo track
mAA at 10 degrees: 0.31569 (±0.00000 over 1 run(s) / ±0.12975 over 9 scenes)
Rank (per category): 76 (of 87)
Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
BM | 1186.0 | — | 278.8 | 0.397 Rank: 84/87 |
0.784 Rank: 81/87 |
0.13775 (±0.00000) Rank: 67/87 |
0.23977 (±0.00000) Rank: 65/87 |
FCS | 1641.0 | — | 480.4 | 0.351 Rank: 77/87 |
0.741 Rank: 79/87 |
0.31222 (±0.00000) Rank: 75/87 |
0.44743 (±0.00000) Rank: 74/87 |
LMS | 780.6 | — | 307.3 | 0.339 Rank: 81/87 |
0.589 Rank: 81/87 |
0.43537 (±0.00000) Rank: 68/87 |
0.57832 (±0.00000) Rank: 64/87 |
LB | 1138.4 | — | 272.6 | 0.324 Rank: 81/87 |
0.596 Rank: 82/87 |
0.22318 (±0.00000) Rank: 67/87 |
0.32730 (±0.00000) Rank: 77/87 |
MC | 1441.5 | — | 315.2 | 0.357 Rank: 82/87 |
0.734 Rank: 82/87 |
0.09013 (±0.00000) Rank: 67/87 |
0.16175 (±0.00000) Rank: 84/87 |
MR | 897.5 | — | 232.3 | 0.375 Rank: 77/87 |
0.786 Rank: 80/87 |
0.08764 (±0.00000) Rank: 83/87 |
0.15065 (±0.00000) Rank: 83/87 |
PSM | 1525.3 | — | 359.5 | 0.278 Rank: 81/87 |
0.493 Rank: 68/87 |
0.12173 (±0.00000) Rank: 54/87 |
0.23761 (±0.00000) Rank: 53/87 |
SF | 1399.7 | — | 447.3 | 0.330 Rank: 79/87 |
0.653 Rank: 82/87 |
0.21836 (±0.00000) Rank: 80/87 |
0.34698 (±0.00000) Rank: 79/87 |
SPC | 1401.2 | — | 357.5 | 0.313 Rank: 84/87 |
0.639 Rank: 80/87 |
0.21615 (±0.00000) Rank: 77/87 |
0.35144 (±0.00000) Rank: 77/87 |
Avg | 1267.9 | — | 339.0 | 0.341 Rank: 82/87 |
0.668 Rank: 80/87 |
0.20473 (±0.00000) Rank: 77/87 |
0.31569 (±0.00000) Rank: 76/87 |
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) / ±0.11868 over 9 scenes)
Rank (per category): 42 (of 87)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
BM | 1186.0 | 281.73 | 100.00 Rank: 1/87 |
1387.57 Rank: 55/87 |
4.569 Rank: 71/87 |
0.37127 Rank: 33/87 |
0.56087 (±0.00000) Rank: 38/87 |
0.70768 (±0.00000) Rank: 35/87 |
FCS | 1641.0 | 484.63 | 97.01 Rank: 59/87 |
2278.96 Rank: 51/87 |
4.586 Rank: 59/87 |
0.27583 Rank: 63/87 |
0.71576 (±0.00000) Rank: 52/87 |
0.77657 (±0.00000) Rank: 54/87 |
LMS | 780.6 | 325.65 | 99.33 Rank: 30/87 |
1123.89 Rank: 73/87 |
5.569 Rank: 34/87 |
0.27147 Rank: 13/87 |
0.84690 (±0.00000) Rank: 28/87 |
0.90388 (±0.00000) Rank: 23/87 |
LB | 1138.4 | 312.59 | 98.36 Rank: 1/87 |
1339.89 Rank: 56/87 |
4.909 Rank: 61/87 |
0.51040 Rank: 66/87 |
0.63548 (±0.00000) Rank: 60/87 |
0.75848 (±0.00000) Rank: 60/87 |
MC | 1441.5 | 312.58 | 99.64 Rank: 59/87 |
1654.19 Rank: 56/87 |
4.487 Rank: 66/87 |
0.36990 Rank: 45/87 |
0.51207 (±0.00000) Rank: 50/87 |
0.66029 (±0.00000) Rank: 52/87 |
MR | 897.5 | 229.21 | 95.00 Rank: 26/87 |
936.07 Rank: 78/87 |
4.787 Rank: 37/87 |
0.55022 Rank: 53/87 |
0.33950 (±0.00000) Rank: 67/87 |
0.46273 (±0.00000) Rank: 66/87 |
PSM | 1525.3 | 359.55 | 99.01 Rank: 37/87 |
2087.99 Rank: 49/87 |
4.182 Rank: 50/87 |
0.31882 Rank: 9/87 |
0.62277 (±0.00000) Rank: 41/87 |
0.71870 (±0.00000) Rank: 38/87 |
SF | 1399.7 | 438.66 | 99.60 Rank: 41/87 |
2181.51 Rank: 56/87 |
4.655 Rank: 59/87 |
0.34509 Rank: 67/87 |
0.73245 (±0.00000) Rank: 47/87 |
0.81715 (±0.00000) Rank: 47/87 |
SPC | 1401.2 | 368.68 | 100.00 Rank: 1/87 |
1775.97 Rank: 57/87 |
4.697 Rank: 58/87 |
0.47533 Rank: 50/87 |
0.72160 (±0.00000) Rank: 42/87 |
0.82736 (±0.00000) Rank: 39/87 |
Avg | 1267.9 | 345.92 | 98.66 Rank: 35/87 |
1640.67 Rank: 58/87 |
4.716 Rank: 58/87 |
0.38759 Rank: 43/87 |
0.63193 (±0.00000) Rank: 43/87 |
0.73698 (±0.00000) Rank: 42/87 |
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: 0.42299 (±0.00000 over 1 run(s) / ±0.05866 over 3 scenes)
Rank (per category): 64 (of 87)
Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
Lizard | 1858.1 | — | 445.4 | 0.031 Rank: 59/87 |
0.011 Rank: 24/87 |
0.30402 (±0.00000) Rank: 63/87 |
0.43051 (±0.00000) Rank: 62/87 |
Pond | 2048.0 | — | 486.3 | 0.046 Rank: 72/87 |
0.042 Rank: 72/87 |
0.32305 (±0.00000) Rank: 64/87 |
0.49078 (±0.00000) Rank: 59/87 |
Tree | 2048.0 | — | 328.2 | 0.027 Rank: 71/87 |
0.019 Rank: 58/87 |
0.21987 (±0.00000) Rank: 68/87 |
0.34768 (±0.00000) Rank: 67/87 |
Avg | 1984.7 | — | 419.9 | 0.034 Rank: 67/87 |
0.024 Rank: 72/87 |
0.28231 (±0.00000) Rank: 67/87 |
0.42299 (±0.00000) Rank: 64/87 |
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) / ±0.14435 over 3 scenes)
Rank (per category): 30 (of 87)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
Lizard | 1858.1 | 465.07 | 85.70 Rank: 2/87 |
1322.90 Rank: 4/87 |
3.176 Rank: 62/87 |
14.57073 Rank: 66/87 |
0.58218 (±0.00000) Rank: 31/87 |
0.67042 (±0.00000) Rank: 26/87 |
Pond | 2048.0 | 493.61 | 68.37 Rank: 34/87 |
1102.56 Rank: 4/87 |
2.748 Rank: 63/87 |
0.43365 Rank: 46/87 |
0.29373 (±0.00000) Rank: 14/87 |
0.31792 (±0.00000) Rank: 26/87 |
Tree | 2048.0 | 293.00 | 71.78 Rank: 25/87 |
553.38 Rank: 28/87 |
2.815 Rank: 66/87 |
5.96140 Rank: 28/87 |
0.42869 (±0.00000) Rank: 37/87 |
0.47028 (±0.00000) Rank: 32/87 |
Avg | 1984.7 | 417.23 | 75.28 Rank: 18/87 |
992.95 Rank: 9/87 |
2.913 Rank: 66/87 |
6.98859 Rank: 59/87 |
0.43487 (±0.00000) Rank: 31/87 |
0.48621 (±0.00000) Rank: 30/87 |
Google Urban dataset / Stereo track
mAA at 10 degrees: 0.16669 (±0.00000 over 1 run(s) / ±0.09324 over 17 scenes)
Rank (per category): 50 (of 87)
Scene | Features | Matches (raw) |
Matches (final) |
mAA(5o) | mAA(10o) |
Amsterdam (AMS) | 1298.5 | — | 209.0 | 0.04554 (±0.00000) Rank: 52/87 |
0.11473 (±0.00000) Rank: 52/87 |
Bangkok (BGK) | 890.4 | — | 219.6 | 0.01935 (±0.00000) Rank: 50/87 |
0.07816 (±0.00000) Rank: 50/87 |
Barcelona (BCN) | 838.7 | — | 164.3 | 0.04835 (±0.00000) Rank: 49/87 |
0.09963 (±0.00000) Rank: 49/87 |
Buenos Aires (BAR) | 906.2 | — | 181.3 | 0.04749 (±0.00000) Rank: 52/87 |
0.12192 (±0.00000) Rank: 52/87 |
Cambridge (CAM) | 860.2 | — | 152.5 | 0.07329 (±0.00000) Rank: 52/87 |
0.16267 (±0.00000) Rank: 50/87 |
Cannes (CAN) | 820.9 | — | 185.5 | 0.14194 (±0.00000) Rank: 51/87 |
0.22811 (±0.00000) Rank: 51/87 |
Chicago (CHI) | 828.7 | — | 120.5 | 0.02664 (±0.00000) Rank: 50/87 |
0.06680 (±0.00000) Rank: 50/87 |
Helsinki (HEL) | 984.3 | — | 288.0 | 0.14588 (±0.00000) Rank: 48/87 |
0.27680 (±0.00000) Rank: 48/87 |
Madrid (MAD) | 882.3 | — | 141.6 | 0.02188 (±0.00000) Rank: 48/87 |
0.08116 (±0.00000) Rank: 49/87 |
Mountain View (MTV) | 779.7 | — | 158.6 | 0.06641 (±0.00000) Rank: 49/87 |
0.13652 (±0.00000) Rank: 49/87 |
New Orleans (NOR) | 707.6 | — | 126.2 | 0.06805 (±0.00000) Rank: 52/87 |
0.14645 (±0.00000) Rank: 50/87 |
San Francisco (SF) | 813.2 | — | 180.7 | 0.04672 (±0.00000) Rank: 51/87 |
0.12073 (±0.00000) Rank: 50/87 |
Singapore (SG) | 854.6 | — | 172.6 | 0.09684 (±0.00000) Rank: 49/87 |
0.18639 (±0.00000) Rank: 49/87 |
Sydney (SYD) | 965.0 | — | 220.5 | 0.12314 (±0.00000) Rank: 49/87 |
0.24672 (±0.00000) Rank: 49/87 |
Tokyo (TOK) | 980.8 | — | 249.0 | 0.29069 (±0.00000) Rank: 49/87 |
0.46152 (±0.00000) Rank: 49/87 |
Toronto (TOR) | 943.2 | — | 177.1 | 0.06960 (±0.00000) Rank: 49/87 |
0.15000 (±0.00000) Rank: 49/87 |
Zurich (ZRH) | 816.3 | — | 200.7 | 0.08247 (±0.00000) Rank: 52/87 |
0.15533 (±0.00000) Rank: 51/87 |
Average (Avg) | 892.4 | — | 185.2 | 0.08319 (±0.00000) Rank: 50/87 |
0.16669 (±0.00000) Rank: 50/87 |
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) / ±0.15571 over 17 scenes)
Rank (per category): 38 (of 87)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
AMS | 1298.5 | 145.32 | 90.40 Rank: 36/87 |
1032.37 Rank: 42/87 |
3.581 Rank: 43/87 |
24.55549 Rank: 20/87 |
0.26739 (±0.00000) Rank: 41/87 |
0.38419 (±0.00000) Rank: 39/87 |
BGK | 890.4 | 160.29 | 96.13 Rank: 34/87 |
1235.14 Rank: 43/87 |
3.577 Rank: 46/87 |
11.45394 Rank: 8/87 |
0.01843 (±0.00000) Rank: 48/87 |
0.06476 (±0.00000) Rank: 47/87 |
BCN | 838.7 | 108.89 | 81.87 Rank: 31/87 |
660.13 Rank: 42/87 |
3.515 Rank: 44/87 |
25.85652 Rank: 21/87 |
0.08201 (±0.00000) Rank: 42/87 |
0.14438 (±0.00000) Rank: 39/87 |
BAR | 906.2 | 117.89 | 84.01 Rank: 32/87 |
771.53 Rank: 38/87 |
3.360 Rank: 30/87 |
16.48927 Rank: 31/87 |
0.14263 (±0.00000) Rank: 34/87 |
0.25001 (±0.00000) Rank: 32/87 |
CAM | 860.2 | 105.79 | 84.48 Rank: 32/87 |
687.44 Rank: 41/87 |
3.285 Rank: 43/87 |
29.11021 Rank: 44/87 |
0.14488 (±0.00000) Rank: 36/87 |
0.22841 (±0.00000) Rank: 34/87 |
CAN | 820.9 | 113.60 | 61.16 Rank: 45/87 |
280.10 Rank: 48/87 |
2.993 Rank: 49/87 |
15.14796 Rank: 14/87 |
0.02590 (±0.00000) Rank: 42/87 |
0.05743 (±0.00000) Rank: 41/87 |
CHI | 828.7 | 108.10 | 89.63 Rank: 17/87 |
386.53 Rank: 33/87 |
4.010 Rank: 38/87 |
47.85968 Rank: 36/87 |
0.03776 (±0.00000) Rank: 43/87 |
0.09956 (±0.00000) Rank: 40/87 |
HEL | 984.3 | 215.70 | 94.80 Rank: 11/87 |
1381.73 Rank: 45/87 |
3.893 Rank: 42/87 |
15.51122 Rank: 43/87 |
0.30022 (±0.00000) Rank: 34/87 |
0.44497 (±0.00000) Rank: 32/87 |
MAD | 882.3 | 104.11 | 88.44 Rank: 36/87 |
867.28 Rank: 39/87 |
3.094 Rank: 47/87 |
5.30349 Rank: 19/87 |
0.06484 (±0.00000) Rank: 37/87 |
0.15204 (±0.00000) Rank: 37/87 |
MTV | 779.7 | 121.65 | 91.39 Rank: 28/87 |
856.15 Rank: 40/87 |
3.650 Rank: 44/87 |
17.29666 Rank: 44/87 |
0.22366 (±0.00000) Rank: 37/87 |
0.34851 (±0.00000) Rank: 37/87 |
NOR | 707.6 | 93.17 | 83.09 Rank: 32/87 |
547.97 Rank: 40/87 |
3.207 Rank: 36/87 |
11.25580 Rank: 19/87 |
0.17102 (±0.00000) Rank: 37/87 |
0.25625 (±0.00000) Rank: 35/87 |
SF | 813.2 | 132.24 | 94.19 Rank: 30/87 |
906.81 Rank: 40/87 |
3.557 Rank: 39/87 |
19.98246 Rank: 26/87 |
0.20138 (±0.00000) Rank: 37/87 |
0.34528 (±0.00000) Rank: 37/87 |
SG | 854.6 | 113.83 | 74.98 Rank: 31/87 |
582.25 Rank: 41/87 |
3.162 Rank: 40/87 |
17.06092 Rank: 27/87 |
0.15728 (±0.00000) Rank: 36/87 |
0.21467 (±0.00000) Rank: 36/87 |
SYD | 965.0 | 149.67 | 90.35 Rank: 15/87 |
940.14 Rank: 38/87 |
3.332 Rank: 39/87 |
17.06526 Rank: 12/87 |
0.21936 (±0.00000) Rank: 39/87 |
0.31777 (±0.00000) Rank: 38/87 |
TOK | 980.8 | 161.81 | 92.00 Rank: 34/87 |
1340.99 Rank: 41/87 |
3.531 Rank: 45/87 |
15.65880 Rank: 50/87 |
0.62230 (±0.00000) Rank: 38/87 |
0.70212 (±0.00000) Rank: 38/87 |
TOR | 943.2 | 125.00 | 86.05 Rank: 36/87 |
1008.38 Rank: 36/87 |
3.086 Rank: 43/87 |
17.38110 Rank: 32/87 |
0.12074 (±0.00000) Rank: 39/87 |
0.23681 (±0.00000) Rank: 39/87 |
ZRH | 816.3 | 136.74 | 86.44 Rank: 32/87 |
1012.58 Rank: 38/87 |
3.279 Rank: 42/87 |
19.76662 Rank: 30/87 |
0.07426 (±0.00000) Rank: 37/87 |
0.13733 (±0.00000) Rank: 37/87 |
Avg | 892.4 | 130.22 | 86.44 Rank: 31/87 |
852.80 Rank: 41/87 |
3.418 Rank: 41/87 |
19.22091 Rank: 21/87 |
0.16906 (±0.00000) Rank: 38/87 |
0.25791 (±0.00000) Rank: 38/87 |
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 —









