Deep Learning for Plastic Waste Classification System

Table 1

Structure of the AlexNet, 23 layers.

Number	Name of layer	Parameters
1	Image input	227 × 227 × 3
2	Convolution	64 filters, size 11 × 11
3	ReLU
4	Cross channel normalization
5	Max pooling
6	Convolution	128 filters, size 5 × 5
7	ReLU
8	Cross channel normalization
9	Max pooling
10	Convolution	128 filters, size 3 × 3
11	ReLU
12	Convolution	192 filters, size 3 × 3
13	ReLU
14	Convolution	128 filters, size 3 × 3
15	ReLU
16	Max pooling
17	Fully connected	Inputs 18432, outputs 512
18	ReLU
19	Fully connected	Inputs 521, outputs 1024
20	ReLU
21	Fully connected	Inputs, outputs 4
22	Soft max
23	Classification	4