Submission ID: 00071
L2-Net w/ DEGENSAC (no FLANN)
Processed: 20-04-23. Download link: sid-00071-l2net-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 (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. 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: 0.52953 (±0.00043 over 3 run(s) / ±0.14975 over 9 scenes)
Rank (per category): 51 (of 147)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 7968.3 | 536.1 | 536.1 | 355.6 | 0.527 Rank: 49/147 |
0.943 Rank: 56/147 |
0.21492 (±0.00099) Rank: 58/147 |
0.33651 (±0.00136) Rank: 56/147 |
| fcs | 7873.6 | 683.9 | 683.9 | 464.8 | 0.443 Rank: 91/147 |
0.943 Rank: 21/147 |
0.62536 (±0.00196) Rank: 41/147 |
0.74425 (±0.00143) Rank: 36/147 |
| lms | 7770.9 | 478.5 | 478.5 | 264.5 | 0.468 Rank: 23/147 |
0.701 Rank: 33/147 |
0.54745 (±0.00007) Rank: 44/147 |
0.66627 (±0.00030) Rank: 45/147 |
| lb | 7952.8 | 452.7 | 452.7 | 271.4 | 0.474 Rank: 92/147 |
0.779 Rank: 32/147 |
0.45987 (±0.00099) Rank: 58/147 |
0.55082 (±0.00068) Rank: 53/147 |
| mc | 7823.8 | 641.3 | 641.3 | 435.7 | 0.524 Rank: 99/147 |
0.959 Rank: 51/147 |
0.38675 (±0.00099) Rank: 58/147 |
0.53645 (±0.00271) Rank: 62/147 |
| mr | 7790.8 | 601.8 | 601.8 | 446.0 | 0.457 Rank: 99/147 |
0.974 Rank: 38/147 |
0.32161 (±0.00113) Rank: 56/147 |
0.43195 (±0.00136) Rank: 60/147 |
| psm | 7807.3 | 304.4 | 304.4 | 180.1 | 0.385 Rank: 102/147 |
0.589 Rank: 82/147 |
0.14919 (±0.00037) Rank: 67/147 |
0.26056 (±0.00081) Rank: 70/147 |
| sf | 7893.6 | 827.2 | 827.2 | 561.6 | 0.479 Rank: 102/147 |
0.902 Rank: 11/147 |
0.50672 (±0.00222) Rank: 51/147 |
0.64416 (±0.00253) Rank: 47/147 |
| spc | 7868.9 | 541.6 | 541.6 | 314.4 | 0.487 Rank: 102/147 |
0.877 Rank: 41/147 |
0.44329 (±0.00205) Rank: 53/147 |
0.59478 (±0.00160) Rank: 53/147 |
| avg | 7861.1 | 563.0 | 563.0 | 366.0 | 0.472 Rank: 103/147 |
0.852 Rank: 44/147 |
0.40613 (±0.00065) Rank: 51/147 |
0.52953 (±0.00043) Rank: 51/147 |
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: 0.66445 (±0.00213 over 3 run(s) / ±0.15616 over 9 scenes)
Rank (per category): 96 (of 147)
| Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 7968.3 | 305.54 | 99.33 Rank: 129/147 |
2257.45 Rank: 120/147 |
4.864 Rank: 99/147 |
0.54205 Rank: 93/147 |
0.40376 (±0.00842) Rank: 89/147 |
0.55428 (±0.00838) Rank: 93/147 |
| fcs | 7873.6 | 428.34 | 97.82 Rank: 84/147 |
4078.05 Rank: 104/147 |
4.378 Rank: 91/147 |
0.25335 Rank: 65/147 |
0.70306 (±0.00528) Rank: 99/147 |
0.77432 (±0.00645) Rank: 96/147 |
| lms | 7770.9 | 278.83 | 98.70 Rank: 97/147 |
2887.34 Rank: 115/147 |
4.317 Rank: 115/147 |
0.27543 Rank: 38/147 |
0.80400 (±0.01065) Rank: 91/147 |
0.86392 (±0.01033) Rank: 90/147 |
| lb | 7952.8 | 320.69 | 98.40 Rank: 65/147 |
2833.02 Rank: 93/147 |
4.500 Rank: 90/147 |
0.51626 Rank: 98/147 |
0.62453 (±0.00223) Rank: 79/147 |
0.72475 (±0.00314) Rank: 79/147 |
| mc | 7823.8 | 364.17 | 99.38 Rank: 102/147 |
3684.43 Rank: 99/147 |
4.379 Rank: 112/147 |
0.41729 Rank: 93/147 |
0.48407 (±0.00684) Rank: 100/147 |
0.63054 (±0.00652) Rank: 104/147 |
| mr | 7790.8 | 411.59 | 92.79 Rank: 111/147 |
4232.90 Rank: 102/147 |
3.985 Rank: 111/147 |
0.54214 Rank: 114/147 |
0.41069 (±0.00608) Rank: 96/147 |
0.51801 (±0.00718) Rank: 97/147 |
| psm | 7807.3 | 122.15 | 82.92 Rank: 119/147 |
1913.19 Rank: 124/147 |
3.129 Rank: 95/147 |
0.74596 Rank: 65/147 |
0.28493 (±0.00821) Rank: 109/147 |
0.34724 (±0.00987) Rank: 111/147 |
| sf | 7893.6 | 536.25 | 97.84 Rank: 76/147 |
5565.18 Rank: 91/147 |
4.237 Rank: 111/147 |
0.34460 Rank: 105/147 |
0.71032 (±0.00295) Rank: 84/147 |
0.79551 (±0.00256) Rank: 83/147 |
| spc | 7868.9 | 283.42 | 98.94 Rank: 77/147 |
3372.91 Rank: 100/147 |
4.063 Rank: 111/147 |
0.43223 Rank: 18/147 |
0.66918 (±0.00566) Rank: 84/147 |
0.77146 (±0.00648) Rank: 85/147 |
| avg | 7861.1 | 339.00 | 96.23 Rank: 113/147 |
3424.94 Rank: 105/147 |
4.206 Rank: 105/147 |
0.45215 Rank: 93/147 |
0.56606 (±0.00238) Rank: 94/147 |
0.66445 (±0.00213) Rank: 96/147 |
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 —
