Table 5.

The quantitative evaluation results of the above network.

Method	Backbone	AP	AP50	AP75	APS	APM	APL
Mask R-CNN	Res-101-FPN	32.67	51.63	34.76	—	10.45	33.52
MS R-CNN	Res-101-FPN	35.21	55.76	38.43	—	11.55	32.51
TensorMask	Res-101-FPN	34.04	56.26	36.36	—	11.26	29.76
PANet	Res-50-FPN	33.54	54.93	36.21	—	11.26	31.23
PolarMask	Res-101-FPN	27.34	48.84	27.94	—	10.54	30.96
YPLACT++	Res-101-FPN	31.53	50.74	33.84	—	14.90	33.23
SOLOv2	Res-101-FPN	34.63	52.66	35.84	—	11.01	46.57
Mask R-CNN (RFN)	Res-101-FPN	37.81	61.51	38.54	—	13.20	39.41
SOLOv2 (RFN)	Res-101-FPN	38.77	62.77	42.14	—	16.44	49.52
Mask R-CNN (∗)	Res-101-FPN	37.97	61.46	39.05	—	15.20	52.70
MSIS	Res-50-FPN	40.06	63.26	45.62	—	17.61	62.37