MAF_HDA: Multi-Object Tracking and Segmentation with Embedding Mask-based Affinity Fusion in Hierarchical Data Association
Video not available.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Rendering of new sequences is currently deactivated due to heavy load.
Benchmark:
MOTS |
Short name:
MAF_HDA
Description:
1st submission: GMPHD filter + KCF (done, Nov 4th, 2021)
2nd submission: GMPHD filter + SiameseRPN (to do, Nov 7th, 2021)
3rd submission: GMPHD filter + DaSiameseRPN (to do, Nov 10th, 2021)
Final submission: best among 1st~3rd submissions (to do, Nov 13th, 2021)
Reference:
Y. Song, Y. Yoon, K. Yoon, M. Jeon. Multi-Object Tracking and Segmentation with Embedding Mask-based Affinity Fusion in Hierarchical Data Association. In IEEE Access, 2022.
Last submitted:
December 01, 2021 (2 years ago)
Published:
December 05, 2021 at 22:26:13 CET
Submissions:
4
Project page / code:
Open source:
Yes
Hardware:
Intel i7-7700K @ 4.2GHz, 4 Cores
Runtime:
4.6 Hz
Benchmark performance:
| Sequence | sMOTSA | IDF1 | MOTSA | MOTSP | MODSA | MT | ML | TP | FP | FN | Rcll | Prcn | ID Sw. | Frag |
| MOTS | 69.9 | 67.0 | 83.8 | 84.1 | 85.1 | 245 (74.7) | 12 (3.7) | 28,311 | 863 | 3,958 | 87.7 | 97.0 | 401 (457.1) | 598 (681.6) |
Detailed performance:
| Sequence | sMOTSA | IDF1 | MOTSA | MOTSP | MODSA | MT | ML | TP | FP | FN | Rcll | Prcn | ID Sw. | Frag |
| MOTS20-01 | 69.9 | 74.2 | 85.2 | 82.5 | 85.4 | 8 | 0 | 2,719 | 65 | 387 | 87.5 | 97.7 | 7 | 44 |
| MOTS20-06 | 74.6 | 66.6 | 88.2 | 85.4 | 90.7 | 153 | 5 | 9,124 | 219 | 690 | 93.0 | 97.7 | 248 | 299 |
| MOTS20-07 | 65.3 | 61.6 | 80.7 | 81.8 | 81.5 | 36 | 4 | 10,874 | 374 | 2,004 | 84.4 | 96.7 | 113 | 189 |
| MOTS20-12 | 71.9 | 75.0 | 82.8 | 87.4 | 83.3 | 48 | 3 | 5,594 | 205 | 877 | 86.4 | 96.5 | 33 | 66 |
Raw data:
