Difference between revisions of "2014:Audio Downbeat Estimation Results"

From MIREX Wiki
(Created page with "Hello!")
 
(→‎Results)
 
(3 intermediate revisions by 2 users not shown)
Line 1: Line 1:
−
Hello!
+
= Submitted Algorithms =
 +
 
 +
{|class="wikitable" style="text-align: left;"
 +
        |+ Algorithms submitted to the Automatic Downbeat Estimation task
 +
! width="80" | Submission code
 +
! width="200" | Submission name
 +
! width="80" style="text-align: center;" | Abstract
 +
! width="440" | Contributors
 +
|-
 +
        !
 +
        !
 +
|-
 +
! DBDR2
 +
| DB1_no_ballroom  ||  style="text-align: center;" | [https://www.music-ir.org/mirex/abstracts/2014/DBDR2.pdf PDF] || Simon Durand, Juan Bello, Bertrand David, Gael Richard
 +
|-
 +
! DBDR3
 +
| DB2_no_beatles  ||  style="text-align: center;" | [https://www.music-ir.org/mirex/abstracts/2014/DBDR3.pdf PDF] || Simon Durand, Juan Bello, Bertrand David, Gael Richard
 +
        |-
 +
! FK3
 +
| Ballroom_tracker  ||  style="text-align: center;" | [https://www.music-ir.org/mirex/abstracts/2014/FK3.pdf PDF] || Florian Krebs
 +
        |-
 +
! FK4
 +
| Ballroom_Pop_tracker  ||  style="text-align: center;" | [https://www.music-ir.org/mirex/abstracts/2014/FK4.pdf PDF] || Florian Krebs
 +
        |-
 +
! KSH1
 +
| Turkish_Indian_Cretan_tracker  ||  style="text-align: center;" | [https://www.music-ir.org/mirex/abstracts/2014/KSH1.pdf PDF] || Florian Krebs, Andre Holzapfel, Ajay Srinivasamurthy
 +
        |-
 +
|}
 +
 
 +
= Results =
 +
 
 +
{| class="wikitable sortable" border="1"
 +
|+ Results ballroom dataset
 +
|-
 +
! Algorithm
 +
! F-Measure
 +
! Precision
 +
! Recall
 +
|-
 +
| FK3*
 +
| 79.4
 +
| 77.7
 +
| 81.8
 +
|-
 +
| FK4*
 +
| 70.7
 +
| 67.8
 +
| 75.8
 +
|-
 +
| KSH1
 +
| 19.5
 +
| 22.9
 +
| 20.0
 +
|-
 +
| DBDR2
 +
| 70.7
 +
| 75.7
 +
| 70.7
 +
|-
 +
| DBDR3*
 +
| 75.4
 +
| 80.3
 +
| 75.4
 +
|}
 +
 
 +
 
 +
{| class="wikitable sortable" border="1"
 +
|+ Results beatles dataset
 +
|-
 +
! Algorithm
 +
! F-Measure
 +
! Precision
 +
! Recall
 +
|-
 +
| FK3
 +
| 58.8
 +
| 57.5
 +
| 63.4
 +
|-
 +
| FK4
 +
| 63.0
 +
| 59.5
 +
| 70.2
 +
|-
 +
| KSH1
 +
| 19.4
 +
| 21.1
 +
| 20.8
 +
|-
 +
| DBDR2*
 +
| 83.1
 +
| 82.6
 +
| 87.1
 +
|-
 +
| DBDR3
 +
| 81.6
 +
| 81.4
 +
| 85.6
 +
|}
 +
 
 +
 
 +
{| class="wikitable sortable" border="1"
 +
|+ Results carnatic dataset
 +
|-
 +
! Algorithm
 +
! F-Measure
 +
! Precision
 +
! Recall
 +
|-
 +
| FK3
 +
| 16.9
 +
| 16.1
 +
| 20.2
 +
|-
 +
| FK4
 +
| 19.4
 +
| 15.3
 +
| 30.4
 +
|-
 +
| KSH1
 +
| 40.0
 +
| 39.9
 +
| 41.4
 +
|-
 +
| DBDR2
 +
| 18.4
 +
| 15.6
 +
| 25.9
 +
|-
 +
| DBDR3
 +
| 20.0
 +
| 16.9
 +
| 28.2
 +
|}
 +
 
 +
 
 +
{| class="wikitable sortable" border="1"
 +
|+ Results turkish dataset
 +
|-
 +
! Algorithm
 +
! F-Measure
 +
! Precision
 +
! Recall
 +
|-
 +
| FK3
 +
| 19.7
 +
| 15.8
 +
| 28.9
 +
|-
 +
| FK4
 +
| 24.0
 +
| 18.2
 +
| 38.7
 +
|-
 +
| KSH1*
 +
| 77.5
 +
| 75.9
 +
| 81.9
 +
|-
 +
| DBDR2
 +
| 44.8
 +
| 40.0
 +
| 55.3
 +
|-
 +
| DBDR3
 +
| 44.8
 +
| 39.5
 +
| 56.5
 +
|}
 +
 
 +
 
 +
{| class="wikitable sortable" border="1"
 +
 
 +
|+ Results cretan dataset
 +
|-
 +
! Algorithm
 +
! F-Measure
 +
! Precision
 +
! Recall
 +
|-
 +
| FK3
 +
| 53.5
 +
| 86.6
 +
| 39.3
 +
|-
 +
| FK4
 +
| 51.2
 +
| 74.5
 +
| 39.4
 +
|-
 +
| KSH1*
 +
| 85.4
 +
| 88.2
 +
| 84.1
 +
|-
 +
| DBDR2
 +
| 43.5
 +
| 65.5
 +
| 32.5
 +
|-
 +
| DBDR3
 +
| 41.5
 +
| 62.5
 +
| 31.0
 +
|}
 +
 
 +
 
 +
{| class="wikitable sortable" border="1"
 +
|+ Results hjdb dataset
 +
|-
 +
! Algorithm
 +
! F-Measure
 +
! Precision
 +
! Recall
 +
|-
 +
| FK3
 +
| 21.5
 +
| 26.2
 +
| 19.1
 +
|-
 +
| FK4
 +
| 64.0
 +
| 64.2
 +
| 63.9
 +
|-
 +
| KSH1
 +
| 9.4
 +
| 16.7
 +
| 8.5
 +
|-
 +
| DBDR2
 +
| 64.2
 +
| 65.1
 +
| 64.2
 +
|-
 +
| DBDR3
 +
| 66.7
 +
| 67.6
 +
| 66.8
 +
|}
 +
 
 +
<nowiki>*</nowiki>) Rows marked by an asterisk should be taken with care as in those cases overlapping test and training sets were used.
 +
 
 +
= Runtime =
 +
All submissions except KSH1 finished the computations in less than 24 hours.

Latest revision as of 13:48, 10 October 2015

Submitted Algorithms

Algorithms submitted to the Automatic Downbeat Estimation task
Submission code Submission name Abstract Contributors
DBDR2 DB1_no_ballroom PDF Simon Durand, Juan Bello, Bertrand David, Gael Richard
DBDR3 DB2_no_beatles PDF Simon Durand, Juan Bello, Bertrand David, Gael Richard
FK3 Ballroom_tracker PDF Florian Krebs
FK4 Ballroom_Pop_tracker PDF Florian Krebs
KSH1 Turkish_Indian_Cretan_tracker PDF Florian Krebs, Andre Holzapfel, Ajay Srinivasamurthy

Results

Results ballroom dataset
Algorithm F-Measure Precision Recall
FK3* 79.4 77.7 81.8
FK4* 70.7 67.8 75.8
KSH1 19.5 22.9 20.0
DBDR2 70.7 75.7 70.7
DBDR3* 75.4 80.3 75.4


Results beatles dataset
Algorithm F-Measure Precision Recall
FK3 58.8 57.5 63.4
FK4 63.0 59.5 70.2
KSH1 19.4 21.1 20.8
DBDR2* 83.1 82.6 87.1
DBDR3 81.6 81.4 85.6


Results carnatic dataset
Algorithm F-Measure Precision Recall
FK3 16.9 16.1 20.2
FK4 19.4 15.3 30.4
KSH1 40.0 39.9 41.4
DBDR2 18.4 15.6 25.9
DBDR3 20.0 16.9 28.2


Results turkish dataset
Algorithm F-Measure Precision Recall
FK3 19.7 15.8 28.9
FK4 24.0 18.2 38.7
KSH1* 77.5 75.9 81.9
DBDR2 44.8 40.0 55.3
DBDR3 44.8 39.5 56.5


Results cretan dataset
Algorithm F-Measure Precision Recall
FK3 53.5 86.6 39.3
FK4 51.2 74.5 39.4
KSH1* 85.4 88.2 84.1
DBDR2 43.5 65.5 32.5
DBDR3 41.5 62.5 31.0


Results hjdb dataset
Algorithm F-Measure Precision Recall
FK3 21.5 26.2 19.1
FK4 64.0 64.2 63.9
KSH1 9.4 16.7 8.5
DBDR2 64.2 65.1 64.2
DBDR3 66.7 67.6 66.8

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

Runtime

All submissions except KSH1 finished the computations in less than 24 hours.

Retrieved from "https://music-ir.org/mirex/w/index.php?title=2014:Audio_Downbeat_Estimation_Results&oldid=11276"