Difference between revisions of "2016:Audio Downbeat Estimation Results"

From MIREX Wiki
(Runtime)
(Runtime)
Line 395: Line 395:
  
 
{| class="wikitable" border="1"
 
{| class="wikitable" border="1"
|+ Runtime
 
 
|-
 
|-
 
! Algorithm
 
! Algorithm

Revision as of 06:44, 28 July 2016

Submitted Algorithms

Algorithms submitted to the Automatic Downbeat Estimation task
Submission code Submission name Abstract Contributors
DBDR1 DB1_no_beatles PDF Simon Durand, Juan Bello, Bertrand David, Gael Richard
DBDR2 DB2_no_ballroom PDF Simon Durand, Juan Bello, Bertrand David, Gael Richard
KB1 babeats13 PDF Florian Krebs, Sebastian Böck
KB2 babeats15 PDF Florian Krebs, Sebastian Böck
BK4 joint_tracker PDF Sebastian Böck, Florian Krebs
DSR1 downbeater PDF Matthew Davies, Adam Stark, Andrew Robertson
CD4 qm-barbeattracker PDF Matthew Davies, Chris Cannam

Results

Results ballroom dataset
Algorithm F-Measure Precision Recall
DBDR1* 0.838 0.874 0.846
DBDR2 0.783 0.808 0.804
BK4* 0.908 0.906 0.917
CD4 0.412 0.416 0.419
DSR1 0.463 0.476 0.468
KB1* 0.898 0.888 0.917
KB2* 0.860 0.853 0.890
Results beatles dataset
Algorithm F-Measure Precision Recall
DBDR1 0.849 0.861 0.868
DBDR2* 0.872 0.861 0.909
BK4* 0.865 0.872 0.876
CD4 0.604 0.586 0.642
DSR1 0.665 0.646 0.708
KB1 0.803 0.776 0.859
KB2* 0.818 0.799 0.870
Results carnatic dataset
Algorithm F-Measure Precision Recall
DBDR1 0.201 0.199 0.240
DBDR2 0.231 0.194 0.330
BK4* 0.369 0.290 0.566
CD4 0.186 0.154 0.258
DSR1 0.184 0.155 0.251
KB1 0.269 0.221 0.380
KB2* 0.330 0.263 0.487
Results turkish dataset
Algorithm F-Measure Precision Recall
DBDR1 0.306 0.292 0.379
DBDR2 0.415 0.360 0.554
BK4* 0.537 0.468 0.729
CD4 0.218 0.186 0.291
DSR1 0.317 0.281 0.411
KB1 0.352 0.301 0.498
KB2* 0.336 0.269 0.513
Results cretan dataset
Algorithm F-Measure Precision Recall
DBDR1 0.426 0.715 0.308
DBDR2 0.418 0.637 0.311
BK4* 0.635 0.951 0.476
CD4 0.250 0.377 0.188
DSR1 0.265 0.398 0.199
KB1 0.433 0.641 0.328
KB2* 0.443 0.661 0.334
Results hjdb dataset
Algorithm F-Measure Precision Recall
DBDR1 0.578 0.613 0.561
DBDR2 0.629 0.628 0.638
BK4* 0.970 0.970 0.970
CD4 0.334 0.341 0.329
DSR1 0.208 0.232 0.196
KB1 0.690 0.693 0.688
KB2* 0.851 0.854 0.848
Results rwc_classical dataset
Algorithm F-Measure Precision Recall
DBDR1* 0.527 0.570 0.529
DBDR2* 0.532 0.539 0.574
BK4* 0.599 0.659 0.598
CD4 0.174 0.189 0.185
DSR1 0.251 0.260 0.279
KB1 0.436 0.475 0.447
KB2* 0.428 0.459 0.444
Results gtzan dataset
Algorithm F-Measure Precision Recall
DBDR1 0.615 0.651 0.631
DBDR2 0.619 0.628 0.666
BK4 0.638 0.636 0.669
CD4 0.460 0.461 0.482
DSR1 0.397 0.397 0.423
KB1 0.630 0.647 0.634
KB2 0.647 0.665 0.653

*) Rows marked by an asterisk should be taken with care as in those cases overlapping test and training sets were used. This could lead to overestimated metrics.

Runtime

The runtime is measured for the *hjdb* dataset (duration 3h19m) and then extrapolated to the duration of all datasets (total 38h51m).

Algorithm Multi-core Runtime hjdb [mm] Extrapolated runtime all [hh] x faster than realtime
DBDRx yes 86.8 16.9 2.29
BK4 yes 13.1 2.55 15.2
DSR1 no 4.5 0.88 44.2
CD4 no 3.1 0.6 63.5