Submission ID: b3d08932
MD-net
Processed: 21-12-29. Download link: b3d08932a4c5b5f6-md-net-2k.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: Anonymous (to be released: 2021-6-18)
- Keypoint: md-net-2k
- Descriptor: md-net-2k (128 float32: 512 bytes)
- Number of features: 2048
- Summary: MD-net
- Paper: N/A
- Website: N/A
- Processing date: 21-12-29
Phototourism dataset / Stereo track
mAA at 10 degrees: 0.46028 (±0.00000 over 1 run(s) / ±0.14766 over 9 scenes)
Rank (per category): 56 (of 87)
| Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| BM | 2046.4 | 354.4 | 239.2 | 0.544 Rank: 18/87 |
0.925 Rank: 22/87 |
0.26168 (±0.00000) Rank: 48/87 |
0.39185 (±0.00000) Rank: 49/87 |
| FCS | 2043.7 | 347.5 | 228.4 | 0.404 Rank: 49/87 |
0.881 Rank: 12/87 |
0.52743 (±0.00000) Rank: 54/87 |
0.66285 (±0.00000) Rank: 54/87 |
| LMS | 2027.8 | 399.0 | 251.4 | 0.394 Rank: 50/87 |
0.700 Rank: 12/87 |
0.56062 (±0.00000) Rank: 36/87 |
0.69636 (±0.00000) Rank: 33/87 |
| LB | 2043.9 | 247.9 | 139.6 | 0.359 Rank: 65/87 |
0.735 Rank: 17/87 |
0.32688 (±0.00000) Rank: 48/87 |
0.44135 (±0.00000) Rank: 62/87 |
| MC | 2037.5 | 344.3 | 242.9 | 0.518 Rank: 18/87 |
0.933 Rank: 6/87 |
0.25571 (±0.00000) Rank: 48/87 |
0.40388 (±0.00000) Rank: 60/87 |
| MR | 2033.1 | 383.7 | 279.6 | 0.494 Rank: 3/87 |
0.943 Rank: 5/87 |
0.23098 (±0.00000) Rank: 45/87 |
0.32811 (±0.00000) Rank: 48/87 |
| PSM | 2027.2 | 271.3 | 171.5 | 0.377 Rank: 5/87 |
0.530 Rank: 61/87 |
0.11490 (±0.00000) Rank: 58/87 |
0.20735 (±0.00000) Rank: 60/87 |
| SF | 2040.2 | 383.8 | 276.1 | 0.441 Rank: 2/87 |
0.837 Rank: 6/87 |
0.42114 (±0.00000) Rank: 52/87 |
0.55888 (±0.00000) Rank: 49/87 |
| SPC | 2028.5 | 279.6 | 165.4 | 0.408 Rank: 26/87 |
0.812 Rank: 22/87 |
0.29949 (±0.00000) Rank: 68/87 |
0.45190 (±0.00000) Rank: 68/87 |
| Avg | 2036.5 | 334.6 | 221.6 | 0.438 Rank: 15/87 |
0.811 Rank: 13/87 |
0.33320 (±0.00000) Rank: 56/87 |
0.46028 (±0.00000) Rank: 56/87 |
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) / ±0.12529 over 9 scenes)
Rank (per category): 59 (of 87)
| Scene | Features | Matches (input) |
Registration Ratio (%) |
Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| BM | 2046.4 | 353.13 | 100.00 Rank: 1/87 |
1087.12 Rank: 67/87 |
5.017 Rank: 27/87 |
0.43419 Rank: 48/87 |
0.51305 (±0.00000) Rank: 47/87 |
0.65379 (±0.00000) Rank: 48/87 |
| FCS | 2043.7 | 358.94 | 97.17 Rank: 52/87 |
1640.05 Rank: 67/87 |
4.493 Rank: 62/87 |
0.25829 Rank: 49/87 |
0.71067 (±0.00000) Rank: 54/87 |
0.77404 (±0.00000) Rank: 58/87 |
| LMS | 2027.8 | 421.33 | 99.27 Rank: 38/87 |
1466.56 Rank: 53/87 |
4.908 Rank: 71/87 |
0.32377 Rank: 64/87 |
0.84626 (±0.00000) Rank: 29/87 |
0.89712 (±0.00000) Rank: 30/87 |
| LB | 2043.9 | 283.04 | 97.63 Rank: 60/87 |
1035.70 Rank: 71/87 |
4.698 Rank: 64/87 |
0.50096 Rank: 57/87 |
0.60657 (±0.00000) Rank: 65/87 |
0.71862 (±0.00000) Rank: 65/87 |
| MC | 2037.5 | 340.16 | 99.87 Rank: 50/87 |
1209.51 Rank: 67/87 |
5.190 Rank: 25/87 |
0.42785 Rank: 64/87 |
0.46788 (±0.00000) Rank: 65/87 |
0.62046 (±0.00000) Rank: 63/87 |
| MR | 2033.1 | 383.88 | 93.40 Rank: 53/87 |
1455.45 Rank: 59/87 |
4.729 Rank: 40/87 |
0.50443 Rank: 25/87 |
0.42842 (±0.00000) Rank: 30/87 |
0.54745 (±0.00000) Rank: 32/87 |
| PSM | 2027.2 | 270.71 | 95.31 Rank: 51/87 |
1874.15 Rank: 58/87 |
3.403 Rank: 61/87 |
0.56281 Rank: 59/87 |
0.38101 (±0.00000) Rank: 63/87 |
0.47269 (±0.00000) Rank: 62/87 |
| SF | 2040.2 | 379.63 | 97.29 Rank: 55/87 |
1711.30 Rank: 67/87 |
4.854 Rank: 50/87 |
0.30161 Rank: 46/87 |
0.71245 (±0.00000) Rank: 54/87 |
0.79591 (±0.00000) Rank: 55/87 |
| SPC | 2028.5 | 286.55 | 99.19 Rank: 57/87 |
1273.73 Rank: 67/87 |
4.547 Rank: 63/87 |
0.43879 Rank: 30/87 |
0.65075 (±0.00000) Rank: 66/87 |
0.76567 (±0.00000) Rank: 65/87 |
| Avg | 2036.5 | 341.93 | 97.68 Rank: 55/87 |
1417.06 Rank: 64/87 |
4.649 Rank: 59/87 |
0.41697 Rank: 52/87 |
0.59078 (±0.00000) Rank: 61/87 |
0.69397 (±0.00000) Rank: 59/87 |
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: 0.57921 (±0.00000 over 1 run(s) / ±0.08056 over 3 scenes)
Rank (per category): 48 (of 87)
| Scene | Features | Matches (raw) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| Lizard | 2048.0 | 296.7 | 163.1 | 0.024 Rank: 73/87 |
0.009 Rank: 47/87 |
0.40499 (±0.00000) Rank: 49/87 |
0.51207 (±0.00000) Rank: 49/87 |
| Pond | 2048.0 | 355.0 | 183.1 | 0.048 Rank: 56/87 |
0.051 Rank: 21/87 |
0.42725 (±0.00000) Rank: 51/87 |
0.53307 (±0.00000) Rank: 55/87 |
| Tree | 2048.0 | 285.0 | 137.2 | 0.028 Rank: 68/87 |
0.021 Rank: 47/87 |
0.58675 (±0.00000) Rank: 45/87 |
0.69249 (±0.00000) Rank: 44/87 |
| Avg | 2048.0 | 312.2 | 161.1 | 0.033 Rank: 71/87 |
0.027 Rank: 30/87 |
0.47300 (±0.00000) Rank: 44/87 |
0.57921 (±0.00000) Rank: 48/87 |
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) / ±0.14130 over 3 scenes)
Rank (per category): 55 (of 87)
| Scene | Features | Matches (input) |
Registration Ratio (%) |
Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| Lizard | 2048.0 | 321.72 | 83.83 Rank: 51/87 |
610.43 Rank: 62/87 |
3.483 Rank: 4/87 |
14.15553 Rank: 31/87 |
0.53987 (±0.00000) Rank: 52/87 |
0.62178 (±0.00000) Rank: 54/87 |
| Pond | 2048.0 | 442.43 | 71.43 Rank: 5/87 |
911.19 Rank: 29/87 |
3.054 Rank: 12/87 |
0.42536 Rank: 32/87 |
0.25384 (±0.00000) Rank: 53/87 |
0.27582 (±0.00000) Rank: 60/87 |
| Tree | 2048.0 | 318.59 | 71.52 Rank: 28/87 |
459.44 Rank: 47/87 |
3.048 Rank: 17/87 |
5.95583 Rank: 27/87 |
0.42076 (±0.00000) Rank: 42/87 |
0.45762 (±0.00000) Rank: 42/87 |
| Avg | 2048.0 | 360.91 | 75.60 Rank: 12/87 |
660.35 Rank: 52/87 |
3.195 Rank: 9/87 |
6.84557 Rank: 26/87 |
0.40482 (±0.00000) Rank: 50/87 |
0.45174 (±0.00000) Rank: 55/87 |
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.
— Lizard —
— Pond —
— Tree —
Google Urban dataset / Stereo track
mAA at 10 degrees: N/A (±N/A over N/A run(s) / ±N/A over 17 scenes)
Rank (per category): (of 86)
No results for this dataset.
Google Urban dataset / Multiview track
mAA at 10 degrees: N/A (±N/A over N/A run(s) / ±N/A over 17 scenes)
Rank (per category): (of 86)
No results for this dataset.