Auxiliary Pneumonia Classification Algorithm Based on Pruning Compression

Figure 2

ResNet50 structure before and after pruning. The left side of the figure is a schematic diagram of the beginning of the original ResNet50 and the structure of the first two bottleneck blocks. The right side is the structure of CPResNet50 after pruning, which represents the pruning rate of the first convolutional layer. Each bottleneck block contains three convolutional layers; a normalization layer follows each convolutional layer. Finally, an activation function introduces nonlinear factors for channel transmission (one more convolutional layer and normalization layer at the Bottleneck).