Submission ID: 00554
ContextDesc-Upright w/ DEGENSAC (no FLANN, single orientation)
Processed: 20-05-11. Download link: sid-00554-contextdesc-upright-8k-both-degensac.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: Zixin Luo, Jiahui Zhang, Hongkai Chen (contact)
- Keypoint: sift-def
- Descriptor: contextdesc-upright (128 float32: 512 bytes)
- Number of features: 8000
- Summary: ContextDesc with 8000 SIFT features, using the built-in matcher (bidirectional filter with the both strategy, optimal inlier and ratio test thresholds) with DEGENSAC, and setting keypoint orientation to a constant value and use only single orientation to increase performance. FLANN disabled.
- Paper: https://arxiv.org/abs/1904.04084
- Website: https://github.com/lzx551402/contextdesc
- Origin: Submission
- Flags: is_submission, is_challenge_2020
Phototourism / Stereo track
mAA at 10 degrees: 0.55697 (±0.00079 over 3 run(s) / ±0.13811 over 9 scenes)
Rank (per category): 28 (of 147)
| Scene | Features | Matches (matcher) |
Matches (filter) |
Matches (final) |
Rep. @ 3 px. | MS @ 3 px. | mAA(5o) | mAA(10o) |
| bm | 7956.0 | 624.1 | 624.1 | 420.1 | 0.521 Rank: 73/147 |
0.951 Rank: 34/147 |
0.26658 (±0.00321) Rank: 25/147 |
0.39992 (±0.00269) Rank: 25/147 |
| fcs | 7838.8 | 794.4 | 794.4 | 558.2 | 0.465 Rank: 19/147 |
0.942 Rank: 24/147 |
0.62765 (±0.00316) Rank: 37/147 |
0.74247 (±0.00146) Rank: 42/147 |
| lms | 7720.6 | 591.1 | 591.1 | 339.1 | 0.467 Rank: 69/147 |
0.705 Rank: 24/147 |
0.56310 (±0.00159) Rank: 28/147 |
0.68056 (±0.00127) Rank: 30/147 |
| lb | 7936.2 | 521.3 | 521.3 | 332.9 | 0.493 Rank: 16/147 |
0.787 Rank: 25/147 |
0.51580 (±0.00321) Rank: 25/147 |
0.61239 (±0.00236) Rank: 26/147 |
| mc | 7779.4 | 830.6 | 830.6 | 585.0 | 0.549 Rank: 21/147 |
0.966 Rank: 15/147 |
0.41952 (±0.00321) Rank: 25/147 |
0.57059 (±0.00230) Rank: 27/147 |
| mr | 7758.6 | 873.1 | 873.1 | 643.3 | 0.479 Rank: 22/147 |
0.978 Rank: 10/147 |
0.34597 (±0.00359) Rank: 24/147 |
0.46142 (±0.00341) Rank: 25/147 |
| psm | 7769.3 | 417.0 | 417.0 | 280.5 | 0.407 Rank: 27/147 |
0.766 Rank: 15/147 |
0.16371 (±0.00143) Rank: 40/147 |
0.28710 (±0.00165) Rank: 39/147 |
| sf | 7861.9 | 1045.6 | 1045.6 | 713.3 | 0.498 Rank: 23/147 |
0.903 Rank: 10/147 |
0.52385 (±0.00170) Rank: 39/147 |
0.65824 (±0.00159) Rank: 38/147 |
| spc | 7850.1 | 584.6 | 584.6 | 363.8 | 0.501 Rank: 31/147 |
0.893 Rank: 9/147 |
0.44855 (±0.00078) Rank: 47/147 |
0.60003 (±0.00087) Rank: 45/147 |
| avg | 7830.1 | 698.0 | 698.0 | 470.7 | 0.487 Rank: 25/147 |
0.877 Rank: 5/147 |
0.43053 (±0.00081) Rank: 28/147 |
0.55697 (±0.00079) Rank: 28/147 |
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: 0.72902 (±0.00255 over 3 run(s) / ±0.11270 over 9 scenes)
Rank (per category): 27 (of 147)
| Scene | Features | Matches (input) |
Registration Ratio (%) |
Number of Landmarks |
Track Length | ATE | mAA(50) | mAA(100) |
| bm | 7956.0 | 453.63 | 99.74 Rank: 73/147 |
3099.65 Rank: 67/147 |
4.876 Rank: 95/147 |
0.44153 Rank: 37/147 |
0.49321 (±0.00570) Rank: 40/147 |
0.64116 (±0.00724) Rank: 36/147 |
| fcs | 7838.8 | 629.21 | 98.52 Rank: 45/147 |
5103.47 Rank: 55/147 |
4.587 Rank: 52/147 |
0.25892 Rank: 81/147 |
0.73553 (±0.00131) Rank: 36/147 |
0.79868 (±0.00466) Rank: 48/147 |
| lms | 7720.6 | 464.33 | 99.44 Rank: 23/147 |
3913.75 Rank: 35/147 |
4.535 Rank: 52/147 |
0.27309 Rank: 31/147 |
0.83361 (±0.01245) Rank: 39/147 |
0.89014 (±0.01224) Rank: 39/147 |
| lb | 7936.2 | 480.34 | 98.63 Rank: 24/147 |
3493.20 Rank: 46/147 |
4.820 Rank: 33/147 |
0.47152 Rank: 18/147 |
0.71338 (±0.00222) Rank: 26/147 |
0.80578 (±0.00268) Rank: 26/147 |
| mc | 7779.4 | 621.34 | 99.74 Rank: 57/147 |
4734.12 Rank: 44/147 |
4.836 Rank: 50/147 |
0.38603 Rank: 39/147 |
0.52120 (±0.00119) Rank: 40/147 |
0.67053 (±0.00064) Rank: 38/147 |
| mr | 7758.6 | 721.83 | 95.06 Rank: 27/147 |
5804.24 Rank: 30/147 |
4.341 Rank: 40/147 |
0.47112 Rank: 20/147 |
0.46007 (±0.01000) Rank: 39/147 |
0.57103 (±0.00877) Rank: 42/147 |
| psm | 7769.3 | 279.12 | 94.69 Rank: 41/147 |
3659.82 Rank: 55/147 |
3.605 Rank: 36/147 |
0.55334 Rank: 20/147 |
0.48277 (±0.01162) Rank: 31/147 |
0.56178 (±0.01053) Rank: 31/147 |
| sf | 7861.9 | 821.12 | 98.50 Rank: 42/147 |
6918.13 Rank: 31/147 |
4.542 Rank: 53/147 |
0.32196 Rank: 53/147 |
0.73585 (±0.00442) Rank: 37/147 |
0.81721 (±0.00578) Rank: 37/147 |
| spc | 7850.1 | 420.15 | 99.33 Rank: 58/147 |
4167.49 Rank: 63/147 |
4.343 Rank: 64/147 |
0.46613 Rank: 52/147 |
0.70571 (±0.00226) Rank: 43/147 |
0.80484 (±0.00255) Rank: 40/147 |
| avg | 7830.1 | 543.45 | 98.18 Rank: 33/147 |
4543.76 Rank: 46/147 |
4.498 Rank: 49/147 |
0.40485 Rank: 34/147 |
0.63126 (±0.00199) Rank: 28/147 |
0.72902 (±0.00255) Rank: 27/147 |
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 —
