Page 176 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications
P. 176
Chapter 6 Plant leaf disease classification based on feature selection 165
The architecture for ResNet illustrated in Figs. 6.9 and 6.10
adds identity mappings known as skip connections. These skip
connections decide the depth of the neural network depending
on the data set. The network first starts training with one residual
block and keeps adding the new layers by checking gradient.
Figure 6.9 Single residual block.
Figure 6.10 ResNet-12 variant architecture.