Skip to main content
. 2021 Jul 2;21(13):4550. doi: 10.3390/s21134550

Table 7.

Average precisions at different thresholds for MI. Bold: First algorithm’s AP 0.9.

AP25 AP30 AP35 AP40 AP45 AP50
V0 A1F0 0.849 0.916 0.947 0.968 0.975 0.979
A1F1 0.610 0.771 0.871 0.923 0.951 0.962
A1F2 0.518 0.679 0.802 0.878 0.913 0.932
A2 0.240 0.329 0.422 0.509 0.589 0.652
V1 A1F0 0.858 0.925 0.952 0.969 0.975 0.978
A1F1 0.638 0.777 0.866 0.910 0.935 0.950
A1F2 0.643 0.749 0.836 0.874 0.902 0.914
A2 0.224 0.299 0.387 0.473 0.552 0.638
V2 A1F0 0.936 0.948 0.963 0.966 0.967 0.969
A1F1 0.812 0.907 0.936 0.951 0.954 0.957
A1F2 0.819 0.906 0.938 0.945 0.952 0.952
A2 0.454 0.542 0.645 0.720 0.777 0.821
V3 A1F0 0.857 0.917 0.947 0.964 0.972 0.976
A1F1 0.577 0.732 0.823 0.877 0.905 0.921
A1F2 0.524 0.674 0.775 0.847 0.888 0.911
A2 0.223 0.310 0.415 0.502 0.580 0.646