Class GoogLeNet

java.lang.Object
ai.djl.basicmodelzoo.cv.classification.GoogLeNet

public final class GoogLeNet extends Object
GoogLeNet uses a stack of a total of 9 inception blocks and global average pooling to generate its estimates. Maximum pooling between inception blocks reduced the dimensionality. The first part is identical to AlexNet and LeNet, the stack of blocks is inherited from VGG and the global average pooling avoids a stack of fully-connected layers at the end.

GoogLeNet paper from Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, Andrew Rabinovich “Going Deeper with Convolutions” https://arxiv.org/abs/1409.4842

See Also:
  • Method Details

    • googLeNet

      public static ai.djl.nn.Block googLeNet(GoogLeNet.Builder builder)
      Creates a GoogLeNet network block with the help of the GoogLeNet Builder.
      Parameters:
      builder - the GoogLeNet.Builder with the necessary arguments.
      Returns:
      a GoogLeNet block.
    • inceptionBlock

      public ai.djl.nn.ParallelBlock inceptionBlock(int c1, int[] c2, int[] c3, int c4)
      Creates a constituent inception block that becomes a part of the whole GoogLeNet model.
      Parameters:
      c1 - number of channels for the first path of sequential block.
      c2 - array of channels for the second path of sequential block.
      c3 - array of channels for the third path of sequential block.
      c4 - number of channels for the fourth path of sequential block.
      Returns:
      a parallel block combining all 4 paths of sequential blocks.
    • builder

      public static GoogLeNet.Builder builder()
      Creates a builder to build a GoogLeNet.
      Returns:
      a new builder