Submission ID: 00136
CV-DoG-AffNet-HardNet-kornia-MAGSAC-2k
Processed: 21-02-05. Download link: sid-00136-dog-affnethardnet-2k-both-magsac.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: sift8k
- Descriptor: affnethardnet (128 float32: 512 bytes)
- Number of features: 2048
- Summary: OpenCV DoG keypoints, followed by AffNet shape estimation and HardNet descriptor. Implementation: OpenCV + kornia library
- Paper: https://arxiv.org/abs/1711.06704
- Website: https://kornia.readthedocs.io/en/latest/feature.html
- Origin: Baseline
- Flags: is_baseline
Phototourism / Stereo track
mAA at 10 degrees: 0.41747 (±0.00072 over 3 run(s) / ±0.14621 over 9 scenes)
Rank (per category): 47 (of 108)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 2048.0 | 373.1 | 373.1 | 294.7 | 0.461 Rank: 30/108 |
0.908 Rank: 63/108 |
0.11337 (±0.00129) Rank: 77/108 |
0.20330 (±0.00164) Rank: 79/108 |
| fcs | 2048.0 | 284.5 | 284.5 | 210.6 | 0.289 Rank: 80/108 |
0.871 Rank: 51/108 |
0.49588 (±0.00048) Rank: 51/108 |
0.62056 (±0.00065) Rank: 55/108 |
| lms | 2048.0 | 346.3 | 346.3 | 246.8 | 0.392 Rank: 26/108 |
0.681 Rank: 57/108 |
0.45906 (±0.00114) Rank: 28/108 |
0.56530 (±0.00128) Rank: 33/108 |
| lb | 2048.0 | 226.1 | 226.1 | 125.6 | 0.311 Rank: 75/108 |
0.673 Rank: 58/108 |
0.31246 (±0.00129) Rank: 77/108 |
0.41259 (±0.00129) Rank: 45/108 |
| mc | 2048.0 | 215.3 | 215.3 | 132.7 | 0.325 Rank: 77/108 |
0.898 Rank: 47/108 |
0.28359 (±0.00129) Rank: 77/108 |
0.41895 (±0.00099) Rank: 40/108 |
| mr | 2046.5 | 296.7 | 296.7 | 226.8 | 0.334 Rank: 47/108 |
0.936 Rank: 28/108 |
0.22258 (±0.00078) Rank: 34/108 |
0.32607 (±0.00106) Rank: 30/108 |
| psm | 2048.0 | 148.4 | 148.4 | 97.3 | 0.274 Rank: 60/108 |
0.395 Rank: 65/108 |
0.09799 (±0.00167) Rank: 44/108 |
0.18369 (±0.00113) Rank: 49/108 |
| sf | 2048.0 | 347.3 | 347.3 | 255.6 | 0.336 Rank: 76/108 |
0.840 Rank: 16/108 |
0.41084 (±0.00229) Rank: 16/108 |
0.54492 (±0.00198) Rank: 19/108 |
| spc | 2048.0 | 253.9 | 253.9 | 166.4 | 0.328 Rank: 47/108 |
0.763 Rank: 77/108 |
0.34298 (±0.00104) Rank: 42/108 |
0.48185 (±0.00105) Rank: 45/108 |
| avg | 2047.8 | 276.8 | 276.8 | 195.2 | 0.339 Rank: 49/108 |
0.774 Rank: 54/108 |
0.30431 (±0.00058) Rank: 43/108 |
0.41747 (±0.00072) Rank: 47/108 |
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: 0.58536 (±0.00125 over 3 run(s) / ±0.19507 over 9 scenes)
Rank (per category): 64 (of 108)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 2048.0 | 370.58 | 99.51 Rank: 71/108 |
1541.76 Rank: 34/108 |
5.258 Rank: 24/108 |
0.72986 Rank: 82/108 |
0.26098 (±0.00505) Rank: 80/108 |
0.39730 (±0.00522) Rank: 80/108 |
| fcs | 2048.0 | 297.20 | 97.84 Rank: 26/108 |
2082.91 Rank: 30/108 |
4.169 Rank: 67/108 |
0.27874 Rank: 52/108 |
0.66667 (±0.00342) Rank: 57/108 |
0.75423 (±0.00516) Rank: 52/108 |
| lms | 2048.0 | 369.35 | 99.24 Rank: 14/108 |
2121.07 Rank: 7/108 |
4.752 Rank: 64/108 |
0.29607 Rank: 21/108 |
0.81167 (±0.00721) Rank: 16/108 |
0.87676 (±0.00639) Rank: 14/108 |
| lb | 2048.0 | 263.96 | 97.17 Rank: 58/108 |
1527.98 Rank: 33/108 |
4.120 Rank: 77/108 |
0.52964 Rank: 70/108 |
0.44481 (±0.01040) Rank: 75/108 |
0.57670 (±0.01052) Rank: 72/108 |
| mc | 2048.0 | 210.23 | 96.45 Rank: 68/108 |
1382.04 Rank: 38/108 |
3.929 Rank: 80/108 |
0.51183 Rank: 72/108 |
0.39147 (±0.00438) Rank: 66/108 |
0.52484 (±0.00583) Rank: 68/108 |
| mr | 2046.5 | 299.55 | 92.18 Rank: 46/108 |
1994.88 Rank: 11/108 |
4.011 Rank: 60/108 |
0.56511 Rank: 59/108 |
0.32980 (±0.00583) Rank: 62/108 |
0.44382 (±0.00814) Rank: 58/108 |
| psm | 2048.0 | 145.18 | 84.51 Rank: 42/108 |
1618.39 Rank: 41/108 |
2.746 Rank: 84/108 |
0.72897 Rank: 60/108 |
0.13258 (±0.00562) Rank: 77/108 |
0.23623 (±0.01246) Rank: 73/108 |
| sf | 2048.0 | 346.38 | 98.76 Rank: 16/108 |
2132.31 Rank: 23/108 |
4.446 Rank: 54/108 |
0.31792 Rank: 22/108 |
0.68768 (±0.00666) Rank: 32/108 |
0.78683 (±0.00955) Rank: 30/108 |
| spc | 2048.0 | 254.10 | 95.96 Rank: 87/108 |
1696.94 Rank: 38/108 |
4.287 Rank: 58/108 |
0.64475 Rank: 86/108 |
0.56862 (±0.00105) Rank: 76/108 |
0.67154 (±0.00193) Rank: 78/108 |
| avg | 2047.8 | 284.06 | 95.74 Rank: 45/108 |
1788.70 Rank: 30/108 |
4.191 Rank: 72/108 |
0.51143 Rank: 66/108 |
0.47714 (±0.00047) Rank: 65/108 |
0.58536 (±0.00125) Rank: 64/108 |
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 —
