Deep Learning for Plastic Waste Classification System

Table 2

Structure of our CNN.

Number	Name of layer	Parameters
1	Image input	120 × 120 x 3
2	Convolution	64 filters, size 9 × 9
3	Max poolingr
4	ReLU r
5	Convolution	64 filters, size 5 × 5
6	ReLU layer
7	Average pooling
8	Convolution	64 filters, size 5 × 5
9	ReLU
10	Average pooling
11	Fully connected	Inputs 10816, outputs 64
12	ReLU
13	Fully connected	Inputs 64, outputs 4
14	Soft max
15	Classification	4