Skip to main content

Table 1 Our architecture of CNN

From: Thermal face recognition under different conditions

Convolution layer-1 (5 * 5, 16)
maxpool  
Convolution layer-2 (3 * 3, 32)
maxpool  
Convolution layer-3 (3 * 3, 46)
maxpool  
Convolution layer-4 (3 * 3, 128)
maxpool  
FC-1024  
FC-1024  
soft-max