Submission ID: 00060
L2-Net w/ MAGSAC
Processed: 20-04-22. Download link: sid-00060-l2net-8k-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: l2net (128 float32: 512 bytes)
- Number of features: 8000
- Summary: 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. MAGSAC for stereo.
- 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: 0.49054 (±0.00043 over 3 run(s) / ±0.15006 over 9 scenes)
Rank (per category): 87 (of 147)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 7968.3 | 495.3 | 495.3 | 414.9 | 0.527 Rank: 49/147 |
0.918 Rank: 108/147 |
0.17425 (±0.00161) Rank: 91/147 |
0.28796 (±0.00271) Rank: 93/147 |
| fcs | 7873.6 | 619.5 | 619.5 | 521.8 | 0.443 Rank: 91/147 |
0.924 Rank: 85/147 |
0.59484 (±0.00478) Rank: 84/147 |
0.72029 (±0.00264) Rank: 82/147 |
| lms | 7770.9 | 431.5 | 431.5 | 334.9 | 0.468 Rank: 23/147 |
0.673 Rank: 95/147 |
0.50682 (±0.00259) Rank: 83/147 |
0.61286 (±0.00123) Rank: 85/147 |
| lb | 7952.8 | 384.8 | 384.8 | 288.9 | 0.474 Rank: 92/147 |
0.729 Rank: 102/147 |
0.37360 (±0.00161) Rank: 91/147 |
0.46783 (±0.00388) Rank: 93/147 |
| mc | 7823.8 | 501.4 | 501.4 | 412.6 | 0.524 Rank: 99/147 |
0.931 Rank: 110/147 |
0.35192 (±0.00161) Rank: 91/147 |
0.50325 (±0.00447) Rank: 98/147 |
| mr | 7790.8 | 437.6 | 437.6 | 394.8 | 0.457 Rank: 99/147 |
0.952 Rank: 97/147 |
0.29430 (±0.00178) Rank: 91/147 |
0.40347 (±0.00092) Rank: 89/147 |
| psm | 7807.3 | 270.6 | 270.6 | 210.9 | 0.385 Rank: 102/147 |
0.490 Rank: 114/147 |
0.13432 (±0.00238) Rank: 85/147 |
0.23758 (±0.00245) Rank: 88/147 |
| sf | 7893.6 | 606.8 | 606.8 | 534.6 | 0.479 Rank: 102/147 |
0.867 Rank: 85/147 |
0.48318 (±0.00401) Rank: 80/147 |
0.61850 (±0.00192) Rank: 77/147 |
| spc | 7868.9 | 477.5 | 477.5 | 367.8 | 0.487 Rank: 102/147 |
0.829 Rank: 107/147 |
0.41112 (±0.00041) Rank: 75/147 |
0.56311 (±0.00142) Rank: 75/147 |
| avg | 7861.1 | 469.5 | 469.5 | 386.8 | 0.472 Rank: 103/147 |
0.813 Rank: 112/147 |
0.36937 (±0.00027) Rank: 90/147 |
0.49054 (±0.00043) Rank: 87/147 |
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: 0.65665 (±0.00287 over 3 run(s) / ±0.15760 over 9 scenes)
Rank (per category): 99 (of 147)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 7968.3 | 297.59 | 99.86 Rank: 39/147 |
2281.55 Rank: 118/147 |
4.876 Rank: 96/147 |
0.52956 Rank: 86/147 |
0.40247 (±0.00960) Rank: 91/147 |
0.55814 (±0.01256) Rank: 89/147 |
| fcs | 7873.6 | 421.57 | 98.23 Rank: 66/147 |
4128.28 Rank: 100/147 |
4.402 Rank: 81/147 |
0.25787 Rank: 77/147 |
0.71092 (±0.00664) Rank: 90/147 |
0.78416 (±0.00585) Rank: 77/147 |
| lms | 7770.9 | 274.27 | 98.26 Rank: 122/147 |
2895.12 Rank: 111/147 |
4.311 Rank: 120/147 |
0.32836 Rank: 120/147 |
0.79334 (±0.00208) Rank: 98/147 |
0.85490 (±0.00386) Rank: 98/147 |
| lb | 7952.8 | 307.22 | 98.22 Rank: 87/147 |
2779.45 Rank: 102/147 |
4.519 Rank: 83/147 |
0.52842 Rank: 110/147 |
0.58809 (±0.00425) Rank: 96/147 |
0.68182 (±0.00543) Rank: 97/147 |
| mc | 7823.8 | 328.15 | 99.08 Rank: 125/147 |
3512.39 Rank: 107/147 |
4.386 Rank: 109/147 |
0.41893 Rank: 96/147 |
0.47939 (±0.00450) Rank: 108/147 |
0.62471 (±0.00396) Rank: 110/147 |
| mr | 7790.8 | 347.46 | 91.92 Rank: 125/147 |
3815.57 Rank: 116/147 |
3.950 Rank: 116/147 |
0.50643 Rank: 83/147 |
0.40354 (±0.00826) Rank: 100/147 |
0.51495 (±0.00850) Rank: 103/147 |
| psm | 7807.3 | 119.22 | 82.54 Rank: 121/147 |
1931.28 Rank: 122/147 |
3.083 Rank: 100/147 |
0.76751 Rank: 66/147 |
0.26523 (±0.00765) Rank: 111/147 |
0.33110 (±0.00834) Rank: 114/147 |
| sf | 7893.6 | 457.27 | 97.43 Rank: 88/147 |
5114.86 Rank: 105/147 |
4.227 Rank: 115/147 |
0.34919 Rank: 110/147 |
0.69776 (±0.00115) Rank: 106/147 |
0.78732 (±0.00145) Rank: 94/147 |
| spc | 7868.9 | 278.12 | 98.87 Rank: 82/147 |
3369.20 Rank: 102/147 |
4.082 Rank: 106/147 |
0.49772 Rank: 83/147 |
0.67066 (±0.00444) Rank: 81/147 |
0.77278 (±0.00444) Rank: 83/147 |
| avg | 7861.1 | 314.54 | 96.05 Rank: 123/147 |
3314.19 Rank: 114/147 |
4.204 Rank: 107/147 |
0.46489 Rank: 108/147 |
0.55682 (±0.00245) Rank: 99/147 |
0.65665 (±0.00287) Rank: 99/147 |
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 —
