Table 1.

Summary of the reviewed architectures for source camera identification. CC and IC denote Constrained and Instant convolutional operations, respectively. GAP denotes the presence of a global average pooling operation. BN denotes the presence of batch normalization operations.

Architecture	Input Size	Preprocessing	Convolutional Part					Fully Connected Part
			N Layers	Activation	Pooling	BN	GAP	N Layers	Activation	Dropout
A1 [23]	48 × 48 × 3	-	3	ReLU	Max	-	-	1	ReLU	✓
A2 [24]	32 × 32 × 3	-	2	L-ReLU	Max	-	-	2	L-ReLU	✓
A3 [25]	36 × 36 × 3	-	3	ReLU	Avg	✓	-	1	ReLU	✓
A4 [26]	64 × 64 × 3	-	13	ReLU	Max	-	-	2	?	✓
A5 [27]	256 × 256 × 3	-	1 Conv, 12 Residual	ReLU	-	-	✓	-	-	-
A6 [36]	64 × 64 × 3	-	4	?	Max	-	-	1	ReLU	-
A7 [37]	64 × 64 × 3	-	10	?	Max	-	-	1	ReLU	-
A8 [38]	256 × 256 × 2	IC + CC	4	TanH	Max, Avg	✓	-	2	TanH	-
A9 [39]	256 × 256	HP	3	ReLU	Max	-	-	2	ReLU	✓
A10 [41]	256 × 256 × 3	LBP	3	ReLU	Max	✓	-	2	ReLU	✓
A11 [42]	64 × 64 × 3	-	6	ReLU	Avg	✓	✓	-	-	-
A12 [43]	64 × 64 × 3	-	1 Conv, 3 Residual	ReLU	Avg	-	✓	-	-	-