Submission ID: ea748742
rss
Processed: 21-06-02. Download link: ea7487426d83bffa-ssa.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: yt-imw2021 (contact)
- Keypoint: spma
- Descriptor: hsg (128 float32: 512 bytes)
- Number of features: 2048
- Summary:
- Paper: N/A
- Website: N/A
- Processing date: 21-06-02
Phototourism dataset / Stereo track
mAA at 10 degrees: 0.58209 (±0.00000 over 1 run(s) / ±0.13489 over 9 scenes)
Rank (per category): 21 (of 87)
Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
BM | 2048.0 | — | 455.2 | 0.449 Rank: 57/87 |
0.885 Rank: 54/87 |
0.32888 (±0.00000) Rank: 26/87 |
0.48143 (±0.00000) Rank: 26/87 |
FCS | 2048.0 | — | 572.0 | 0.405 Rank: 47/87 |
0.869 Rank: 38/87 |
0.66091 (±0.00000) Rank: 6/87 |
0.77867 (±0.00000) Rank: 6/87 |
LMS | 2048.0 | — | 542.8 | 0.420 Rank: 36/87 |
0.679 Rank: 47/87 |
0.61423 (±0.00000) Rank: 19/87 |
0.73826 (±0.00000) Rank: 20/87 |
LB | 2048.0 | — | 416.9 | 0.392 Rank: 47/87 |
0.691 Rank: 50/87 |
0.48633 (±0.00000) Rank: 26/87 |
0.60271 (±0.00000) Rank: 29/87 |
MC | 2048.0 | — | 486.5 | 0.412 Rank: 52/87 |
0.848 Rank: 59/87 |
0.35210 (±0.00000) Rank: 26/87 |
0.50619 (±0.00000) Rank: 25/87 |
MR | 2048.0 | — | 450.6 | 0.375 Rank: 78/87 |
0.886 Rank: 50/87 |
0.27977 (±0.00000) Rank: 26/87 |
0.39352 (±0.00000) Rank: 27/87 |
PSM | 2048.0 | — | 418.3 | 0.328 Rank: 59/87 |
0.616 Rank: 27/87 |
0.24074 (±0.00000) Rank: 20/87 |
0.39920 (±0.00000) Rank: 25/87 |
SF | 2048.0 | — | 519.7 | 0.349 Rank: 66/87 |
0.776 Rank: 53/87 |
0.51472 (±0.00000) Rank: 20/87 |
0.66106 (±0.00000) Rank: 16/87 |
SPC | 2048.0 | — | 466.2 | 0.375 Rank: 57/87 |
0.793 Rank: 51/87 |
0.50937 (±0.00000) Rank: 26/87 |
0.67780 (±0.00000) Rank: 26/87 |
Avg | 2048.0 | — | 480.9 | 0.389 Rank: 63/87 |
0.782 Rank: 47/87 |
0.44300 (±0.00000) Rank: 25/87 |
0.58209 (±0.00000) Rank: 21/87 |
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) / ±0.11141 over 9 scenes)
Rank (per category): 24 (of 87)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
BM | 2048.0 | 455.81 | 99.89 Rank: 29/87 |
2389.86 Rank: 6/87 |
4.959 Rank: 32/87 |
0.35472 Rank: 25/87 |
0.58345 (±0.00000) Rank: 27/87 |
0.72040 (±0.00000) Rank: 28/87 |
FCS | 2048.0 | 580.53 | 98.20 Rank: 26/87 |
2869.20 Rank: 7/87 |
5.077 Rank: 32/87 |
0.22983 Rank: 13/87 |
0.76288 (±0.00000) Rank: 6/87 |
0.81478 (±0.00000) Rank: 12/87 |
LMS | 2048.0 | 568.19 | 99.27 Rank: 38/87 |
2197.56 Rank: 17/87 |
5.542 Rank: 38/87 |
0.27225 Rank: 14/87 |
0.86167 (±0.00000) Rank: 16/87 |
0.90996 (±0.00000) Rank: 18/87 |
LB | 2048.0 | 491.53 | 98.36 Rank: 1/87 |
2285.04 Rank: 10/87 |
5.256 Rank: 35/87 |
0.50045 Rank: 56/87 |
0.72517 (±0.00000) Rank: 18/87 |
0.81543 (±0.00000) Rank: 16/87 |
MC | 2048.0 | 478.31 | 100.00 Rank: 1/87 |
2456.77 Rank: 18/87 |
5.170 Rank: 27/87 |
0.34810 Rank: 18/87 |
0.56626 (±0.00000) Rank: 12/87 |
0.71254 (±0.00000) Rank: 10/87 |
MR | 2048.0 | 442.52 | 92.47 Rank: 63/87 |
2476.02 Rank: 18/87 |
4.556 Rank: 60/87 |
0.51261 Rank: 30/87 |
0.40042 (±0.00000) Rank: 42/87 |
0.51144 (±0.00000) Rank: 44/87 |
PSM | 2048.0 | 416.01 | 99.33 Rank: 25/87 |
2699.25 Rank: 16/87 |
4.619 Rank: 11/87 |
0.35458 Rank: 32/87 |
0.65818 (±0.00000) Rank: 9/87 |
0.74290 (±0.00000) Rank: 12/87 |
SF | 2048.0 | 511.89 | 99.80 Rank: 27/87 |
2990.38 Rank: 6/87 |
4.903 Rank: 42/87 |
0.29360 Rank: 41/87 |
0.77863 (±0.00000) Rank: 26/87 |
0.86112 (±0.00000) Rank: 25/87 |
SPC | 2048.0 | 477.17 | 100.00 Rank: 1/87 |
2523.28 Rank: 13/87 |
5.172 Rank: 28/87 |
0.44068 Rank: 32/87 |
0.75951 (±0.00000) Rank: 15/87 |
0.85003 (±0.00000) Rank: 19/87 |
Avg | 2048.0 | 491.33 | 98.59 Rank: 40/87 |
2543.04 Rank: 7/87 |
5.028 Rank: 35/87 |
0.36742 Rank: 24/87 |
0.67735 (±0.00000) Rank: 18/87 |
0.77096 (±0.00000) Rank: 24/87 |
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: 0.76028 (±0.00000 over 1 run(s) / ±0.05277 over 3 scenes)
Rank (per category): 15 (of 87)
Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
Lizard | 2048.0 | — | 395.8 | 0.037 Rank: 19/87 |
0.011 Rank: 23/87 |
0.58835 (±0.00000) Rank: 15/87 |
0.72150 (±0.00000) Rank: 21/87 |
Pond | 2048.0 | — | 413.2 | 0.052 Rank: 27/87 |
0.047 Rank: 38/87 |
0.58798 (±0.00000) Rank: 22/87 |
0.72445 (±0.00000) Rank: 15/87 |
Tree | 2048.0 | — | 283.5 | 0.032 Rank: 24/87 |
0.021 Rank: 51/87 |
0.73333 (±0.00000) Rank: 12/87 |
0.83488 (±0.00000) Rank: 14/87 |
Avg | 2048.0 | — | 364.2 | 0.040 Rank: 21/87 |
0.026 Rank: 52/87 |
0.63655 (±0.00000) Rank: 15/87 |
0.76028 (±0.00000) Rank: 15/87 |
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) / ±0.12934 over 3 scenes)
Rank (per category): 13 (of 87)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
Lizard | 2048.0 | 418.92 | 83.97 Rank: 50/87 |
1277.25 Rank: 15/87 |
3.338 Rank: 53/87 |
14.17621 Rank: 35/87 |
0.56502 (±0.00000) Rank: 42/87 |
0.64419 (±0.00000) Rank: 47/87 |
Pond | 2048.0 | 440.37 | 66.70 Rank: 58/87 |
1050.89 Rank: 16/87 |
2.803 Rank: 57/87 |
0.41598 Rank: 14/87 |
0.29329 (±0.00000) Rank: 15/87 |
0.32829 (±0.00000) Rank: 10/87 |
Tree | 2048.0 | 267.40 | 72.99 Rank: 3/87 |
673.35 Rank: 5/87 |
3.053 Rank: 14/87 |
6.08920 Rank: 45/87 |
0.46936 (±0.00000) Rank: 7/87 |
0.50706 (±0.00000) Rank: 7/87 |
Avg | 2048.0 | 375.56 | 74.55 Rank: 48/87 |
1000.50 Rank: 7/87 |
3.065 Rank: 40/87 |
6.89380 Rank: 37/87 |
0.44256 (±0.00000) Rank: 15/87 |
0.49318 (±0.00000) Rank: 13/87 |
Google Urban dataset / Stereo track
mAA at 10 degrees: 0.40856 (±0.00000 over 1 run(s) / ±0.13356 over 17 scenes)
Rank (per category): 4 (of 87)
Scene | Features | Matches (raw) |
Matches (final) |
mAA(5o) | mAA(10o) |
Amsterdam (AMS) | 2048.0 | — | 288.8 | 0.25000 (±0.00000) Rank: 19/87 |
0.38437 (±0.00000) Rank: 26/87 |
Bangkok (BGK) | 2046.2 | — | 425.1 | 0.07395 (±0.00000) Rank: 5/87 |
0.18462 (±0.00000) Rank: 8/87 |
Barcelona (BCN) | 2048.0 | — | 308.1 | 0.13407 (±0.00000) Rank: 8/87 |
0.22015 (±0.00000) Rank: 8/87 |
Buenos Aires (BAR) | 2048.0 | — | 292.1 | 0.22192 (±0.00000) Rank: 19/87 |
0.37397 (±0.00000) Rank: 20/87 |
Cambridge (CAM) | 2048.0 | — | 296.1 | 0.31849 (±0.00000) Rank: 19/87 |
0.47603 (±0.00000) Rank: 9/87 |
Cannes (CAN) | 2048.0 | — | 356.8 | 0.42488 (±0.00000) Rank: 1/87 |
0.54424 (±0.00000) Rank: 1/87 |
Chicago (CHI) | 2048.0 | — | 240.2 | 0.19891 (±0.00000) Rank: 24/87 |
0.33183 (±0.00000) Rank: 20/87 |
Helsinki (HEL) | 2048.0 | — | 551.3 | 0.37784 (±0.00000) Rank: 21/87 |
0.57268 (±0.00000) Rank: 24/87 |
Madrid (MAD) | 2048.0 | — | 212.0 | 0.10274 (±0.00000) Rank: 1/87 |
0.22705 (±0.00000) Rank: 2/87 |
Mountain View (MTV) | 2048.0 | — | 267.9 | 0.25508 (±0.00000) Rank: 8/87 |
0.39746 (±0.00000) Rank: 10/87 |
New Orleans (NOR) | 2048.0 | — | 226.9 | 0.28817 (±0.00000) Rank: 9/87 |
0.39852 (±0.00000) Rank: 9/87 |
San Francisco (SF) | 2048.0 | — | 341.2 | 0.21470 (±0.00000) Rank: 7/87 |
0.37008 (±0.00000) Rank: 6/87 |
Singapore (SG) | 2048.0 | — | 274.2 | 0.34747 (±0.00000) Rank: 9/87 |
0.46835 (±0.00000) Rank: 14/87 |
Sydney (SYD) | 2048.0 | — | 325.0 | 0.32664 (±0.00000) Rank: 25/87 |
0.49039 (±0.00000) Rank: 22/87 |
Tokyo (TOK) | 2048.0 | — | 378.3 | 0.61127 (±0.00000) Rank: 9/87 |
0.74534 (±0.00000) Rank: 10/87 |
Toronto (TOR) | 2048.0 | — | 232.3 | 0.20880 (±0.00000) Rank: 26/87 |
0.37480 (±0.00000) Rank: 24/87 |
Zurich (ZRH) | 2048.0 | — | 348.2 | 0.29553 (±0.00000) Rank: 1/87 |
0.38557 (±0.00000) Rank: 1/87 |
Average (Avg) | 2047.9 | — | 315.6 | 0.27356 (±0.00000) Rank: 4/87 |
0.40856 (±0.00000) Rank: 4/87 |
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) / ±0.15344 over 17 scenes)
Rank (per category): 3 (of 87)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
AMS | 2048.0 | 191.81 | 92.67 Rank: 14/87 |
1453.66 Rank: 15/87 |
3.823 Rank: 17/87 |
27.35549 Rank: 46/87 |
0.38602 (±0.00000) Rank: 5/87 |
0.50577 (±0.00000) Rank: 9/87 |
BGK | 2046.2 | 296.46 | 96.67 Rank: 25/87 |
2687.01 Rank: 9/87 |
3.610 Rank: 38/87 |
13.80869 Rank: 33/87 |
0.03137 (±0.00000) Rank: 9/87 |
0.09818 (±0.00000) Rank: 7/87 |
BCN | 2048.0 | 184.63 | 85.81 Rank: 17/87 |
1323.15 Rank: 13/87 |
3.826 Rank: 13/87 |
28.51327 Rank: 37/87 |
0.13628 (±0.00000) Rank: 22/87 |
0.21234 (±0.00000) Rank: 25/87 |
BAR | 2048.0 | 172.01 | 85.42 Rank: 28/87 |
1310.12 Rank: 10/87 |
3.470 Rank: 8/87 |
16.95717 Rank: 44/87 |
0.17173 (±0.00000) Rank: 13/87 |
0.27754 (±0.00000) Rank: 14/87 |
CAM | 2048.0 | 187.68 | 91.47 Rank: 1/87 |
1387.98 Rank: 10/87 |
3.631 Rank: 7/87 |
29.62573 Rank: 49/87 |
0.22278 (±0.00000) Rank: 3/87 |
0.32429 (±0.00000) Rank: 2/87 |
CAN | 2048.0 | 198.57 | 77.13 Rank: 27/87 |
871.83 Rank: 23/87 |
3.563 Rank: 15/87 |
19.59068 Rank: 28/87 |
0.07396 (±0.00000) Rank: 20/87 |
0.11693 (±0.00000) Rank: 28/87 |
CHI | 2048.0 | 213.10 | 92.40 Rank: 3/87 |
727.50 Rank: 6/87 |
4.804 Rank: 20/87 |
48.34189 Rank: 41/87 |
0.16540 (±0.00000) Rank: 8/87 |
0.26858 (±0.00000) Rank: 7/87 |
HEL | 2048.0 | 394.40 | 94.40 Rank: 21/87 |
2461.31 Rank: 4/87 |
4.112 Rank: 4/87 |
15.79863 Rank: 48/87 |
0.34912 (±0.00000) Rank: 17/87 |
0.47827 (±0.00000) Rank: 20/87 |
MAD | 2048.0 | 136.69 | 91.41 Rank: 14/87 |
1527.71 Rank: 6/87 |
3.290 Rank: 22/87 |
7.40988 Rank: 45/87 |
0.11004 (±0.00000) Rank: 5/87 |
0.22064 (±0.00000) Rank: 10/87 |
MTV | 2048.0 | 194.91 | 91.47 Rank: 26/87 |
1576.81 Rank: 7/87 |
3.801 Rank: 33/87 |
15.98096 Rank: 23/87 |
0.27760 (±0.00000) Rank: 22/87 |
0.40656 (±0.00000) Rank: 25/87 |
NOR | 2048.0 | 147.55 | 86.70 Rank: 27/87 |
1254.13 Rank: 17/87 |
3.347 Rank: 27/87 |
13.90358 Rank: 42/87 |
0.28131 (±0.00000) Rank: 19/87 |
0.37244 (±0.00000) Rank: 19/87 |
SF | 2048.0 | 225.08 | 95.73 Rank: 22/87 |
1809.09 Rank: 9/87 |
3.705 Rank: 26/87 |
19.42871 Rank: 21/87 |
0.30221 (±0.00000) Rank: 11/87 |
0.44363 (±0.00000) Rank: 11/87 |
SG | 2048.0 | 170.92 | 79.31 Rank: 13/87 |
1183.84 Rank: 17/87 |
3.236 Rank: 35/87 |
19.37119 Rank: 46/87 |
0.22390 (±0.00000) Rank: 24/87 |
0.29563 (±0.00000) Rank: 24/87 |
SYD | 2048.0 | 202.40 | 93.68 Rank: 1/87 |
1771.07 Rank: 3/87 |
3.615 Rank: 15/87 |
24.84859 Rank: 52/87 |
0.35356 (±0.00000) Rank: 5/87 |
0.45154 (±0.00000) Rank: 6/87 |
TOK | 2048.0 | 238.60 | 92.93 Rank: 22/87 |
2182.49 Rank: 5/87 |
3.769 Rank: 16/87 |
14.33127 Rank: 23/87 |
0.67670 (±0.00000) Rank: 10/87 |
0.74747 (±0.00000) Rank: 12/87 |
TOR | 2048.0 | 145.66 | 89.33 Rank: 12/87 |
1478.31 Rank: 13/87 |
3.249 Rank: 15/87 |
18.72467 Rank: 46/87 |
0.21575 (±0.00000) Rank: 3/87 |
0.35957 (±0.00000) Rank: 1/87 |
ZRH | 2048.0 | 217.44 | 91.25 Rank: 7/87 |
1899.01 Rank: 4/87 |
3.518 Rank: 10/87 |
21.67761 Rank: 42/87 |
0.16010 (±0.00000) Rank: 3/87 |
0.24687 (±0.00000) Rank: 1/87 |
Avg | 2047.9 | 206.93 | 89.87 Rank: 15/87 |
1582.65 Rank: 5/87 |
3.669 Rank: 16/87 |
20.92165 Rank: 51/87 |
0.24340 (±0.00000) Rank: 3/87 |
0.34272 (±0.00000) Rank: 3/87 |
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 —









