Fig. 6. Structure of the densely connected P3D CNN.

The network consists of convolutional layers, a series of dense blocks, a fully connected layer and a softmax layer. As shown in the inset, each dense spatio-temporal convolution block was constructed by introducing skip connections between the input and output of the convolutional layers in the channel dimension, where red represents the input of the dense block, green and blue represent the output of the spatial and temporal convolutional layers, respectively, and yellow represents the output of the entire block