Submission ID: 00611
sift and hardnet64 train scale(12)
Processed: 20-06-02. Download link: sid-00611-sift8k_8000_hardnet64-train-all-l2-val-14000.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: Ximin Zheng, Sheng He, Hualong Shi (contact)
- Keypoint: sift8k
- Descriptor: hardnet64-train-all-l2-val-14000 (128 float32: 512 bytes)
- Number of features: 8000
- Summary: SIFT with 8000 keypoints(scale 12), hardnet64 with 128 descriptors(trained with l2 loss and step 14000), FLANN disabled
- Paper: N/A
- Website: N/A
- Origin: Submission
- Flags: is_submission, is_challenge_2020
Phototourism / Stereo track
mAA at 10 degrees: 0.58870 (±0.00041 over 3 run(s) / ±0.13740 over 9 scenes)
Rank (per category): 5 (of 147)
Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
bm | 7956.1 | 1028.4 | 1028.4 | 685.6 | 0.520 Rank: 81/147 |
0.952 Rank: 25/147 |
0.30113 (±0.00024) Rank: 10/147 |
0.43205 (±0.00043) Rank: 11/147 |
fcs | 7838.8 | 1142.2 | 1142.2 | 779.4 | 0.465 Rank: 25/147 |
0.912 Rank: 99/147 |
0.63978 (±0.00219) Rank: 18/147 |
0.74670 (±0.00063) Rank: 29/147 |
lms | 7720.4 | 714.7 | 714.7 | 391.6 | 0.467 Rank: 108/147 |
0.698 Rank: 44/147 |
0.60205 (±0.00266) Rank: 3/147 |
0.72377 (±0.00167) Rank: 6/147 |
lb | 7936.1 | 677.7 | 677.7 | 418.2 | 0.492 Rank: 55/147 |
0.799 Rank: 10/147 |
0.58235 (±0.00024) Rank: 10/147 |
0.68072 (±0.00268) Rank: 2/147 |
mc | 7779.6 | 1104.7 | 1104.7 | 755.7 | 0.549 Rank: 27/147 |
0.966 Rank: 23/147 |
0.44237 (±0.00024) Rank: 10/147 |
0.59368 (±0.00254) Rank: 4/147 |
mr | 7758.6 | 1080.4 | 1080.4 | 772.9 | 0.479 Rank: 63/147 |
0.977 Rank: 20/147 |
0.36252 (±0.00130) Rank: 5/147 |
0.48145 (±0.00094) Rank: 3/147 |
psm | 7769.4 | 562.1 | 562.1 | 369.3 | 0.407 Rank: 31/147 |
0.771 Rank: 13/147 |
0.18831 (±0.00179) Rank: 17/147 |
0.32042 (±0.00262) Rank: 17/147 |
sf | 7861.7 | 1283.9 | 1283.9 | 857.6 | 0.498 Rank: 32/147 |
0.873 Rank: 72/147 |
0.54178 (±0.00111) Rank: 18/147 |
0.66660 (±0.00071) Rank: 24/147 |
spc | 7850.0 | 914.5 | 914.5 | 568.8 | 0.501 Rank: 37/147 |
0.887 Rank: 18/147 |
0.50472 (±0.00151) Rank: 9/147 |
0.65291 (±0.00057) Rank: 8/147 |
avg | 7830.1 | 945.4 | 945.4 | 622.1 | 0.486 Rank: 33/147 |
0.871 Rank: 17/147 |
0.46278 (±0.00082) Rank: 5/147 |
0.58870 (±0.00041) Rank: 5/147 |
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: 0.75127 (±0.00234 over 3 run(s) / ±0.10107 over 9 scenes)
Rank (per category): 14 (of 147)
Scene | Features | Matches (input) |
RegistrationRatio (%) | Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
bm | 7956.1 | 954.25 | 99.82 Rank: 46/147 |
5049.87 Rank: 11/147 |
5.050 Rank: 35/147 |
0.37951 Rank: 5/147 |
0.55863 (±0.00177) Rank: 6/147 |
0.69893 (±0.00021) Rank: 6/147 |
fcs | 7838.8 | 1107.77 | 97.34 Rank: 109/147 |
7189.21 Rank: 12/147 |
4.707 Rank: 19/147 |
0.25271 Rank: 60/147 |
0.72589 (±0.00767) Rank: 67/147 |
0.78213 (±0.00756) Rank: 84/147 |
lms | 7720.4 | 697.44 | 99.51 Rank: 14/147 |
4979.79 Rank: 13/147 |
4.637 Rank: 13/147 |
0.27345 Rank: 35/147 |
0.85646 (±0.01153) Rank: 2/147 |
0.90814 (±0.01081) Rank: 3/147 |
lb | 7936.1 | 757.56 | 98.70 Rank: 9/147 |
4570.87 Rank: 16/147 |
4.850 Rank: 18/147 |
0.47111 Rank: 17/147 |
0.74724 (±0.00329) Rank: 13/147 |
0.82910 (±0.00376) Rank: 14/147 |
mc | 7779.6 | 1014.66 | 99.94 Rank: 24/147 |
6044.36 Rank: 9/147 |
5.003 Rank: 13/147 |
0.36117 Rank: 19/147 |
0.54490 (±0.00522) Rank: 18/147 |
0.68745 (±0.00503) Rank: 18/147 |
mr | 7758.6 | 1028.89 | 95.08 Rank: 26/147 |
6932.63 Rank: 7/147 |
4.448 Rank: 13/147 |
0.46336 Rank: 7/147 |
0.49288 (±0.00481) Rank: 5/147 |
0.60372 (±0.00481) Rank: 7/147 |
psm | 7769.4 | 485.73 | 97.95 Rank: 17/147 |
5625.21 Rank: 17/147 |
3.669 Rank: 16/147 |
0.47390 Rank: 8/147 |
0.52913 (±0.01082) Rank: 10/147 |
0.60565 (±0.01045) Rank: 13/147 |
sf | 7861.7 | 1184.40 | 97.92 Rank: 69/147 |
8284.72 Rank: 9/147 |
4.743 Rank: 14/147 |
0.30888 Rank: 29/147 |
0.73512 (±0.00190) Rank: 39/147 |
0.81370 (±0.00173) Rank: 45/147 |
spc | 7850.0 | 861.55 | 99.49 Rank: 51/147 |
6098.83 Rank: 12/147 |
4.714 Rank: 10/147 |
0.48840 Rank: 75/147 |
0.74146 (±0.00336) Rank: 15/147 |
0.83264 (±0.00410) Rank: 17/147 |
avg | 7830.1 | 899.14 | 98.42 Rank: 23/147 |
6086.16 Rank: 12/147 |
4.647 Rank: 10/147 |
0.38583 Rank: 15/147 |
0.65908 (±0.00224) Rank: 13/147 |
0.75127 (±0.00234) Rank: 14/147 |
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 —