Submission ID: 00072
L2-Net-Upright w/ DEGENSAC (no FLANN)
Processed: 20-04-22. Download link: sid-00072-l2net-upright-8k-both-degensac-no-flann.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: l2net-upright (128 float32: 512 bytes)
- Number of features: 8000
- Summary: Upright L2-Net descriptors on DoG features (OpenCV), extracted with a low detection threshold to generate up to 8000 points. Using the built-in matcher: bidirectional filter with the both strategy. DEGENSAC for stereo. FLANN disabled.
- Paper: http://openaccess.thecvf.com/content_cvpr_2017/papers/Tian_L2-Net_Deep_Learning_CVPR_2017_paper.pdf
- Website: https://github.com/yuruntian/L2-Net
- Origin: Baseline
- Flags: is_baseline
Phototourism / Stereo track
mAA at 10 degrees: 0.54497 (±0.00003 over 3 run(s) / ±0.13861 over 9 scenes)
Rank (per category): 36 (of 147)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 7956.3 | 541.4 | 541.4 | 355.3 | 0.520 Rank: 93/147 |
0.947 Rank: 42/147 |
0.25806 (±0.00143) Rank: 29/147 |
0.38291 (±0.00108) Rank: 27/147 |
| fcs | 7838.2 | 778.5 | 778.5 | 535.1 | 0.465 Rank: 40/147 |
0.927 Rank: 78/147 |
0.61484 (±0.00096) Rank: 56/147 |
0.72962 (±0.00122) Rank: 67/147 |
| lms | 7720.4 | 518.3 | 518.3 | 286.1 | 0.467 Rank: 74/147 |
0.695 Rank: 59/147 |
0.56059 (±0.00228) Rank: 31/147 |
0.68434 (±0.00267) Rank: 29/147 |
| lb | 7935.2 | 519.5 | 519.5 | 319.9 | 0.492 Rank: 25/147 |
0.779 Rank: 31/147 |
0.50208 (±0.00143) Rank: 29/147 |
0.60087 (±0.00030) Rank: 31/147 |
| mc | 7779.6 | 794.8 | 794.8 | 553.6 | 0.549 Rank: 42/147 |
0.963 Rank: 36/147 |
0.40507 (±0.00143) Rank: 29/147 |
0.55660 (±0.00165) Rank: 35/147 |
| mr | 7758.9 | 824.4 | 824.4 | 607.6 | 0.479 Rank: 30/147 |
0.976 Rank: 30/147 |
0.34227 (±0.00213) Rank: 30/147 |
0.45577 (±0.00203) Rank: 30/147 |
| psm | 7768.4 | 372.8 | 372.8 | 242.6 | 0.407 Rank: 44/147 |
0.697 Rank: 35/147 |
0.15769 (±0.00123) Rank: 52/147 |
0.27749 (±0.00080) Rank: 52/147 |
| sf | 7860.0 | 1032.2 | 1032.2 | 706.0 | 0.498 Rank: 42/147 |
0.877 Rank: 64/147 |
0.51522 (±0.00188) Rank: 45/147 |
0.64241 (±0.00120) Rank: 48/147 |
| spc | 7849.5 | 533.7 | 533.7 | 315.0 | 0.501 Rank: 51/147 |
0.871 Rank: 50/147 |
0.42826 (±0.00135) Rank: 67/147 |
0.57469 (±0.00078) Rank: 69/147 |
| avg | 7829.6 | 657.3 | 657.3 | 435.7 | 0.486 Rank: 45/147 |
0.859 Rank: 31/147 |
0.42045 (±0.00039) Rank: 35/147 |
0.54497 (±0.00003) Rank: 36/147 |
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: 0.68491 (±0.00338 over 3 run(s) / ±0.13083 over 9 scenes)
Rank (per category): 81 (of 147)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 7956.3 | 302.15 | 99.84 Rank: 43/147 |
2116.91 Rank: 125/147 |
4.912 Rank: 85/147 |
0.48585 Rank: 66/147 |
0.45229 (±0.00571) Rank: 60/147 |
0.60108 (±0.00648) Rank: 63/147 |
| fcs | 7838.2 | 470.98 | 96.64 Rank: 128/147 |
4036.14 Rank: 108/147 |
4.570 Rank: 54/147 |
0.25797 Rank: 78/147 |
0.69314 (±0.01178) Rank: 109/147 |
0.76760 (±0.01276) Rank: 104/147 |
| lms | 7720.4 | 297.00 | 98.60 Rank: 102/147 |
2834.02 Rank: 118/147 |
4.434 Rank: 80/147 |
0.29962 Rank: 90/147 |
0.82194 (±0.00256) Rank: 62/147 |
0.87993 (±0.00363) Rank: 58/147 |
| lb | 7935.2 | 367.58 | 98.20 Rank: 91/147 |
2951.73 Rank: 81/147 |
4.684 Rank: 57/147 |
0.50285 Rank: 83/147 |
0.66029 (±0.00696) Rank: 62/147 |
0.75770 (±0.00929) Rank: 63/147 |
| mc | 7779.6 | 452.30 | 99.68 Rank: 67/147 |
3990.37 Rank: 80/147 |
4.666 Rank: 68/147 |
0.37998 Rank: 36/147 |
0.50064 (±0.00535) Rank: 74/147 |
0.64922 (±0.00564) Rank: 75/147 |
| mr | 7758.9 | 567.51 | 92.49 Rank: 114/147 |
4866.95 Rank: 65/147 |
4.149 Rank: 72/147 |
0.48120 Rank: 39/147 |
0.43868 (±0.00712) Rank: 60/147 |
0.54714 (±0.00964) Rank: 68/147 |
| psm | 7768.4 | 162.72 | 87.25 Rank: 107/147 |
2406.04 Rank: 102/147 |
3.428 Rank: 56/147 |
0.66107 Rank: 47/147 |
0.36201 (±0.00582) Rank: 85/147 |
0.43279 (±0.00583) Rank: 92/147 |
| sf | 7860.0 | 663.44 | 96.18 Rank: 125/147 |
6139.30 Rank: 69/147 |
4.427 Rank: 65/147 |
0.32852 Rank: 75/147 |
0.69840 (±0.00255) Rank: 100/147 |
0.77882 (±0.00319) Rank: 108/147 |
| spc | 7849.5 | 276.08 | 97.30 Rank: 128/147 |
3093.16 Rank: 109/147 |
4.165 Rank: 98/147 |
0.67455 Rank: 134/147 |
0.64922 (±0.00756) Rank: 92/147 |
0.74989 (±0.00670) Rank: 94/147 |
| avg | 7829.6 | 395.53 | 96.24 Rank: 111/147 |
3603.85 Rank: 88/147 |
4.382 Rank: 64/147 |
0.45240 Rank: 94/147 |
0.58629 (±0.00295) Rank: 81/147 |
0.68491 (±0.00338) Rank: 81/147 |
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 —
