A Vortex Identification Method Based on Extreme Learning Machine

Table 2

The performance of different methods in the triangle flow field of and the square flow field of .

Cases	Methods	Precision (%)	Recall (%)	Training time	Execution time (s)
Triangle	-criterion	89.15	42.8	\	1.7
	-criterion	87.7	38.4	\	2.1
	-criterion	89.1	41.4	\	2.5
	-criterion	88.9	41.35	\	4.3
	Random Forest	92.9	71.1	112 s	12.7
	AdaBoost	85.5	54.3	150 s	18.14
	Vortex-CNN	95.85	88.2	>24 h	19.45
	Vortex-Seg-Net	96.64	94.72	>24 h	0.81
	Vortex-ELM-Net	96.17	89.4	90.72 s	2.4
	IVD	100	100	\	227
Square	-criterion	93.22	61.6	\	3.6
	-criterion	89.7	60.4	\	4.4
	-criterion	93.1	56.7	\	5.3
	-criterion	93.1	60.7	\	8.6
	Random Forest	87.35	90.1	112 s	24.6
	AdaBoost	88.2	83.44	150 s	33.2
	Vortex-CNN	84.4	95.7	>24 h	41.4
	Vortex-Seg-Net	86.65	90.98	>24 h	1.24
	Vortex-ELM-Net	92.6	91.5	90.72 s	3.9
	IVD	100	100	\	433