Submission ID: 00062
L2-Net-Upright w/ MAGSAC
Processed: 20-04-22. Download link: sid-00062-l2net-upright-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-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. 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: 0.51450 (±0.00058 over 3 run(s) / ±0.14032 over 9 scenes)
Rank (per category): 68 (of 147)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 7956.3 | 512.2 | 512.2 | 431.3 | 0.520 Rank: 93/147 |
0.923 Rank: 101/147 |
0.20907 (±0.00118) Rank: 68/147 |
0.33222 (±0.00138) Rank: 60/147 |
| fcs | 7838.2 | 709.0 | 709.0 | 615.9 | 0.465 Rank: 40/147 |
0.906 Rank: 104/147 |
0.59694 (±0.00194) Rank: 81/147 |
0.71311 (±0.00208) Rank: 86/147 |
| lms | 7720.4 | 475.9 | 475.9 | 380.8 | 0.467 Rank: 74/147 |
0.663 Rank: 107/147 |
0.53546 (±0.00336) Rank: 56/147 |
0.64760 (±0.00250) Rank: 61/147 |
| lb | 7935.2 | 460.7 | 460.7 | 358.3 | 0.492 Rank: 25/147 |
0.740 Rank: 92/147 |
0.44549 (±0.00118) Rank: 68/147 |
0.54654 (±0.00280) Rank: 57/147 |
| mc | 7779.6 | 630.5 | 630.5 | 541.9 | 0.549 Rank: 42/147 |
0.936 Rank: 103/147 |
0.37035 (±0.00118) Rank: 68/147 |
0.52450 (±0.00236) Rank: 79/147 |
| mr | 7758.9 | 618.9 | 618.9 | 568.8 | 0.479 Rank: 30/147 |
0.953 Rank: 93/147 |
0.31992 (±0.00437) Rank: 61/147 |
0.43061 (±0.00521) Rank: 63/147 |
| psm | 7768.4 | 329.5 | 329.5 | 269.7 | 0.407 Rank: 44/147 |
0.610 Rank: 73/147 |
0.14596 (±0.00178) Rank: 72/147 |
0.25967 (±0.00267) Rank: 73/147 |
| sf | 7860.0 | 766.7 | 766.7 | 692.8 | 0.498 Rank: 42/147 |
0.844 Rank: 109/147 |
0.49938 (±0.00070) Rank: 56/147 |
0.62605 (±0.00061) Rank: 64/147 |
| spc | 7849.5 | 483.6 | 483.6 | 385.1 | 0.501 Rank: 51/147 |
0.821 Rank: 110/147 |
0.40268 (±0.00088) Rank: 84/147 |
0.55024 (±0.00034) Rank: 84/147 |
| avg | 7829.6 | 554.1 | 554.1 | 471.6 | 0.486 Rank: 45/147 |
0.822 Rank: 103/147 |
0.39169 (±0.00019) Rank: 68/147 |
0.51450 (±0.00058) Rank: 68/147 |
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: 0.68108 (±0.00177 over 3 run(s) / ±0.13051 over 9 scenes)
Rank (per category): 85 (of 147)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 7956.3 | 297.26 | 99.48 Rank: 115/147 |
2139.35 Rank: 123/147 |
4.877 Rank: 93/147 |
0.48733 Rank: 67/147 |
0.44154 (±0.00325) Rank: 75/147 |
0.59192 (±0.00375) Rank: 74/147 |
| fcs | 7838.2 | 466.96 | 97.32 Rank: 112/147 |
4054.33 Rank: 106/147 |
4.640 Rank: 38/147 |
0.26573 Rank: 96/147 |
0.69785 (±0.00415) Rank: 104/147 |
0.77122 (±0.00490) Rank: 101/147 |
| lms | 7720.4 | 297.33 | 98.43 Rank: 115/147 |
2887.49 Rank: 113/147 |
4.458 Rank: 71/147 |
0.27166 Rank: 29/147 |
0.81802 (±0.00544) Rank: 66/147 |
0.87582 (±0.00463) Rank: 71/147 |
| lb | 7935.2 | 358.47 | 98.11 Rank: 98/147 |
2923.64 Rank: 84/147 |
4.729 Rank: 52/147 |
0.50154 Rank: 78/147 |
0.63996 (±0.00749) Rank: 73/147 |
0.73690 (±0.00899) Rank: 72/147 |
| mc | 7779.6 | 410.92 | 99.36 Rank: 104/147 |
3870.96 Rank: 87/147 |
4.678 Rank: 65/147 |
0.40903 Rank: 79/147 |
0.50379 (±0.00449) Rank: 69/147 |
0.65078 (±0.00374) Rank: 73/147 |
| mr | 7758.9 | 489.60 | 94.67 Rank: 51/147 |
4659.67 Rank: 75/147 |
4.255 Rank: 59/147 |
0.50127 Rank: 73/147 |
0.43649 (±0.00462) Rank: 66/147 |
0.54894 (±0.00558) Rank: 65/147 |
| psm | 7768.4 | 157.61 | 87.32 Rank: 105/147 |
2429.49 Rank: 100/147 |
3.402 Rank: 60/147 |
0.67012 Rank: 48/147 |
0.35785 (±0.01292) Rank: 88/147 |
0.42879 (±0.01590) Rank: 94/147 |
| sf | 7860.0 | 574.44 | 96.12 Rank: 130/147 |
5765.95 Rank: 77/147 |
4.449 Rank: 63/147 |
0.32978 Rank: 78/147 |
0.69936 (±0.00300) Rank: 98/147 |
0.78324 (±0.00337) Rank: 103/147 |
| spc | 7849.5 | 273.78 | 97.01 Rank: 131/147 |
3119.28 Rank: 107/147 |
4.175 Rank: 96/147 |
0.59929 Rank: 126/147 |
0.64099 (±0.00202) Rank: 106/147 |
0.74212 (±0.00115) Rank: 103/147 |
| avg | 7829.6 | 369.60 | 96.42 Rank: 105/147 |
3538.91 Rank: 97/147 |
4.407 Rank: 60/147 |
0.44842 Rank: 91/147 |
0.58176 (±0.00197) Rank: 84/147 |
0.68108 (±0.00177) Rank: 85/147 |
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 —
