Table 1.

Detailed parameters of the ADCN model

Layer name	Kernel size	Stride	Padding	Output size
Data	-	-	-	96*96*29*1
Denseblock1(Db1)_Dilated Conv	3*3*1*32	1*1*1	SAME	96*96*29*32
Db1_Dilated Conv	3*3*1*32	1*1*1	SAME	96*96*29*32
Db1_sum	-	-	-	96*96*29*32
Db1_Dilated Conv	3*3*1*32	1*1*1	SAME	96*96*29*32
Db1_Dilated Conv	3*3*1*32	1*1*1	SAME	96*96*29*32
Db1_sum	-	-	-	96*96*29*32
Fuse	1*1*3	1*1*1	VALID	96*96*27*32
Down sampling	3*3*1	2*2*1	SAME	48*48*27*32
Denseblock2(Db2)_Dilated Conv	3*3*1*32	1*1*1	SAME	48*48*27*32
Db2_Dilated Conv	3*3*1*32	1*1*1	SAME	48*48*27*32
Db2_sum	-	-	-	48*48*27*32
Db2_Dilated Conv	3*3*1*32	1*1*1	SAME	48*48*27*32
Db2_Dilated Conv	3*3*1*32	1*1*1	SAME	48*48*27*32
Db2_sum	-	-	-	48*48*27*32
Fuse	1*1*3	1*1*1	VALID	48*48*25*32
Down sampling	3*3*1	2*2*1	SAME	48*48*25*32
Denseblock3(Db3)_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*25*32
Db3_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*25*32
Db3_sum	-	-	-	24*24*25*32
Db3_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*25*32
Db3_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*25*32
Db3_sum	-	-	-	24*24*25*32
Db3_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*25*32
Db3_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*25*32
Db3_sum	-	-	-	24*24*25*32
Fuse	1*1*3	1*1*1	VALID	24*24*23*32
Denseblock4(Db4)_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*23*32
Db4_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*23*32
Db4_sum	-	-	-	24*24*23*32
Db4_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*23*32
Db4_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*23*32
Db4_sum	-	-	-	24*24*23*32
Db4_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*23*32
Db4_Dilated Conv	3*3*1*32	1*1*1	SAME	24*24*23*32
Db4_sum	-	-	-	24*24*23*32
Fuse	1*1*3	1*1*1	VALID	24*24*23*32
Up sampling1	-	-	-	96*96*21*2
Up sampling2	-	-	-	96*96*21*4
Up sampling3	-	-	-	96*96*21*8
Concat	-	-	-	96*96*21*14
Conv	3*3*1*2	1*1*1	SAME	96*96*21*2