Vehicle Type Recognition Algorithm Based on Improved Network in Network

Algorithm 1

Partial source code of optimized NIN.

	Input:
	input_shape: Input shape of network, default as (224,224,3)
	nclass: Numbers of class (output shape of network), default as 1000
	Output: Optimized NIN model
	The optimized NIN model is established according to the following steps:
	Step 1: Build two residual blocks including 384 convolution kernels
	Build two convolution layers
	Build Max pool layer
	Step 2: Build two residual blocks including 384 convolution kernels
	Build two convolution layers
	Build Max pool layer
	Step 3: Build residual block including 384 convolution kernels
	Build two convolution layers
	Build Max pool layer
	Step 4: Build residual block including 1024 convolution kernels
	Build two convolution layers
	Step 5: Build GAP layers
	return model