Aircraft Detection for Remote Sensing Image Based on Bidirectional and Dense Feature Fusion

Table 3

The contrastive results in the NWPU-DataSet.

AP (%)	Method
	EB-V-F-BR [45]	AD-FCN [46]	CPISNet [47]	RICNN [44]	FRCN [12]	R-P-F RCN [48]	NEOON [49]	DConvNet [37]	DNN [50]	BDFFDN (ours)
Airplane	47.85	69.38	43.10	88.35	82.80	90.40	78.29	87.30	92.40	99.02
Ship	35.41	61.82	58.20	77.34	77.50	75.00	81.68	81.40	79.30	78.89
Storage tank	63.52	69.66	74.60	85.27	52.50	44.40	94.62	63.60	87.10	90.67
Baseball diamond	42.91	62.58	86.20	88.12	96.30	89.90	89.74	90.40	93.20	90.68
Tennis court	52.47	61.23	74.50	40.83	62.90	79.70	61.25	81.60	81.00	90.91
Basketball court	55.57	73.21	83.60	58.45	68.80	77.60	65.04	74.10	89.30	81.50
Ground track field	47.47	75.28	92.50	86.73	98.40	87.70	93.23	90.30	75.80	100.00
Harbor	39.85	57.83	66.60	68.60	82.50	79.10	73.15	75.30	72.50	90.70
Bridge	36.83	53.77	35.70	61.51	78.80	68.20	59.46	71.40	72.80	86.23
Vehicle	37.26	56.38	59.70	71.10	63.80	73.20	78.26	75.50	83.00	81.30
mAP	45.91	64.11	67.50	72.63	76.40	76.50	77.50	79.10	82.60	88.99