Table 2.

Training configuration of the dataset.

Item	AISD	MBD	WHU
Input size	2611 × 2453	1500 × 1500	512 × 512
Train size	512 × 512	512 × 512	512 × 512
Test size	512 × 512	512 × 512	512 × 512
Freeze epoch	50	50	50
UnFreeze epoch	500	600	200
Batch size	8	16	4
Optimizer	Sgd	Sgd	Sgd
Learning decay	Cos	Cos	Cos
Weight decay	0.0005	0.0005	0.0005
Learning rate	1.00 × 10−5	1.00 × 10−5	1.00 × 10−5