A Ghost Bottleneck is a specific type of skip connection block used in the GhostNet CNN architecture. Similar to the basic residual block in ResNet, it integrates several convolutional layers and shortcuts. However, instead of integrating basic residual blocks, the Ghost Bottleneck stacks Ghost Modules instead.

The Ghost Module Structure

The Ghost Module structure consists of two stacked Ghost modules. The first module acts as an expansion layer, increasing the number of channels. The ratio between the number of output channels and that of the input is known as the expansion ratio. The second Ghost module reduces the number of channels to match the shortcut path. The shortcut is connected between the inputs and the outputs of these two Ghost modules.

Batch Normalization and ReLU Nonlinearity

Batch normalization (BN) and ReLU nonlinearity are applied after each layer, except for the second Ghost module where ReLU is not used. This was suggested by MobileNetV2.

The Ghost Bottleneck with a Stride of 1

For a stride of 1, the Ghost bottleneck is described as follows:

  1. The first Ghost module acts as an expansion layer, increasing the number of channels.
  2. The second Ghost module reduces the number of channels to match the shortcut path.
  3. The shortcut is connected between the inputs and the outputs of these two Ghost modules.
  4. Batch normalization and ReLU nonlinearity are applied after each layer, except for the second Ghost module where ReLU is not used.

The Ghost Bottleneck with a Stride of 2

For a stride of 2, the Ghost bottleneck is implemented by a downsampling layer and a depthwise convolution with stride=2 between the two Ghost modules.

The Primary Convolution

In practice, the primary convolution in Ghost module is a pointwise convolution for its efficiency.

Overall, the Ghost Bottleneck is a useful structure in the GhostNet CNN architecture. It allows for efficient processing and improved accuracy in image classification tasks.

Great! Next, complete checkout for full access to SERP AI.
Welcome back! You've successfully signed in.
You've successfully subscribed to SERP AI.
Success! Your account is fully activated, you now have access to all content.
Success! Your billing info has been updated.
Your billing was not updated.