Deepfake Video Detection Based on EfficientNet-V2 Network

Table 2

EfficientNet-B0 network structure.

Stage	Operator	Resolution	Channels	Layers
1	Conv3 ∗ 3	224 ∗ 224	32	1
2	MBConv1, k3 ∗ 3	112 ∗ 112	16	1
3	MBConv6, k3 ∗ 3	112 ∗ 112	24	2
4	MBConv6, k3 ∗ 3	56 ∗ 56	40	2
5	MBConv6, k3 ∗ 3	28 ∗ 28	80	3
6	MBConv6, k5 ∗ 5	14 ∗ 14	112	3
7	MBConv6, k5 ∗ 5	14 ∗ 14	192	4
8	MBConv6, k3 ∗ 3	7 ∗ 7	320	1
9	Conv1 ∗ 1&Pooling&FC	7 ∗ 7	1280	1