VOC2006 PRELIMINARY RESULTS

Key to abbreviations

AbbreviationEmail/submission
AP06_Batradbatra@andrew.cmu.edu
AP06_Leedlee1@andrew.cmu.edu
Cambridgejamieshotton@gmail.com
ENSMPFabien.Moutarde@ensmp.fr
INRIA_Douzematthijs.douze@inrialpes.fr
INRIA_Laptevivan.laptev@inria.fr
INRIA_Larlusdiane.larlus@inrialpes.fr
INRIA_Marszalekmarcin.marszalek@inrialpes.fr
INRIA_Moosmannfrank.moosmann@inrialpes.fr/round1
INRIA_Nowakeric.nowak@inrialpes.fr
INSARouenalain.rakotomamonjy@insa-rouen.fr
KULathomas@esat.kuleuven.be
MIT_Fergusfergus@csail.mit.edu
MIT_Torralbatorralba@csail.mit.edu
MUL_1vALLmartin.antenreiter@unileoben.ac.at/OneVsAllLPU2Ny
MUL_1v1martin.antenreiter@unileoben.ac.at/OneVsOne_SVML
QMUL_HSLSjgzhang@dcs.qmul.ac.uk/HS+LS+SVM
QMUL_LSPCHjgzhang@dcs.qmul.ac.uk/LS+SpatialPyramid+SVM
RWTH_DiscHistdeselaers@informatik.rwth-aachen.de/RWTH-DiscriminativePatchHistograms
RWTH_GMMdeselaers@informatik.rwth-aachen.de/RWTH-GaussianMixtures
RWTH_SparseHistsdeselaers@informatik.rwth-aachen.de/RWTH-SparsePatchHistograms
Sienagabrimonfa@gmail.com/round1
TUDfritz@mis.tu-darmstadt.de
TKKvviitani@cc.hut.fi/round1/PICSOM_ALL
TKKvviitani@cc.hut.fi/round1/PICSOM_DET
UVA_big5jvgemert@science.uva.nl/big5
UVA_weibulljvgemert@science.uva.nl/weibull
XRCEFlorent.Perronnin@xrce.xerox.com
ROUND2_INRIA_Moosmannfrank.moosmann@inrialpes.fr/round2
ROUND2_Sienagabrimonfa@gmail.com/round2
ROUND2_TKKvviitani@cc.hut.fi/round2

Classification results

Competition "comp1" (train on VOC data)

Area under curve (AUC)

 bicycle bus car cat cow dog horse motorbike person sheep
AP06_Batra 0.791 0.637 0.833 0.733 0.756 0.644 0.607 0.672 0.550 0.792
AP06_Lee 0.845 0.916 0.897 0.859 0.838 0.766 0.694 0.829 0.622 0.875
Cambridge 0.873 0.864 0.887 0.822 0.850 0.768 0.754 0.844 0.715 0.866
INRIA_Larlus 0.903 0.948 0.943 0.870 0.880 0.743 0.850 0.890 0.736 0.892
INRIA_Marszalek 0.929 0.984 0.971 0.922 0.938 0.856 0.908 0.964 0.845 0.944
INRIA_Moosmann 0.903 0.933 0.957 0.883 0.895 0.825 0.824 - 0.780 0.930
INRIA_Nowak 0.924 0.973 0.971 0.906 0.892 0.797 0.904 0.961 0.814 0.940
INSARouen - - 0.895 - - 0.764 - - - 0.869
MUL_1vALL 0.857 0.852 0.914 0.562 0.632 0.584 0.525 0.831 0.616 0.758
MUL_1v1 0.864 0.945 0.928 0.826 0.789 0.764 0.733 0.906 0.718 0.872
QMUL_HSLS 0.944 0.984 0.977 0.936 0.936 0.874 0.922 0.966 0.845 0.946
QMUL_LSPCH 0.948 0.981 0.975 0.937 0.938 0.876 0.926 0.969 0.855 0.956
RWTH_DiscHist 0.874 0.955 0.930 0.879 0.910 0.799 0.854 0.938 0.764 0.906
RWTH_GMM 0.882 0.935 0.942 0.866 0.856 0.825 0.802 0.905 0.718 0.892
RWTH_SparseHists 0.863 0.941 0.935 0.883 0.883 0.704 0.844 0.858 0.776 0.907
Siena 0.671 0.749 0.842 0.696 0.774 0.677 0.644 0.701 0.660 0.768
TKK 0.857 0.928 0.943 0.871 0.892 0.811 0.806 0.908 0.781 0.900
UVA_big5 0.897 0.929 0.945 0.845 0.862 0.785 0.806 0.923 0.774 0.885
UVA_weibull 0.855 0.880 0.910 0.818 0.849 0.762 0.759 0.888 0.723 0.811
XRCE 0.943 0.978 0.967 0.933 0.940 0.866 0.925 0.957 0.863 0.951
  Second round submissions
ROUND2_INRIA_Moosmann 0.924 0.962 0.972 0.893 0.925 0.851 0.877 0.934 0.840 0.942
ROUND2_Siena - - - - 0.699 - - - - -
ROUND2_TKK 0.843 0.940 0.951 0.876 0.898 0.835 0.856 0.908 0.778 0.899

ROC curves

Equal error rate (EER) analysis

Images by median ranking across all methods

Competition "comp2" (train on own data)

Area under curve (AUC)

 bicycle bus car cat cow dog horse motorbike person sheep
KUL - - - - - - - 0.797 - -
MIT_Fergus - - 0.763 - - - - 0.821 - -
MIT_Torralba - - 0.745 - - - - - - -

ROC curves

Equal error rate (EER) analysis

Detection results

Competition "comp3" (train on VOC data)

Average precision (AP)

 bicycle bus car cat cow dog horse motorbike person sheep
Cambridge 0.249 0.138 0.254 0.151 0.149 0.118 0.091 0.178 0.030 0.131
ENSMP - - 0.398 - 0.159 - - - - -
INRIA_Douze 0.414 0.117 0.444 - 0.212 - - 0.390 0.164 0.251
INRIA_Laptev 0.440 - - - 0.224 - 0.140 0.318 0.114 -
TUD - - - - - - - 0.153 0.074 -
TKK 0.303 0.169 0.222 0.160 0.252 0.113 0.137 0.265 0.039 0.227

PR curves

Competition "comp4" (train on own data)

Average precision (AP)

 bicycle bus car cat cow dog horse motorbike person sheep
INRIA_Douze - - - - - - - - 0.162 -
KUL - - - - - - - 0.229 - -
MIT_Fergus - - 0.160 - - - - 0.159 - -
MIT_Torralba - - 0.217 - - - - - - -

PR curves