Table 5.

Proposed Resnet152V2+Bi-GRU architecture.

Layer (type)	Output Shape	Parameters
Resnet152v2	(None, 7, 7, 2048)	58331648
reshape (Reshape)	(None, 7, 7, 2048)	0
Time_distributed_7	(None, 7, 14336)	0
bidirectional GRU	(None, 512)	22416384
dropout (Dropout)	(None, 512)	0
dense (Dense)	(None, 4)	2052
Total params: 80,750,084
Trainable params: 80,606,340
Non-trainable params: 143,744