Skip to content

conv2

2D convolution.

This layer performs 2D convolution operations on an input.


Parameters

  • strides : Integer(s), stride values.
    Two integers mean [stride_y, stride_x].
    One integer means stride_y and stride_x have the same value.
    The default is 1.
  • dilations : Integer(s), dilation ratios.
    Two integers mean [dilation_y, dilation_x].
    One integer means dilation_y and dilation_x have the same value.
    The default is 1.
  • padding : String, padding mode.
    "same_upper" and "valid" are compatible with "same" and "valid"on Keras (TensorFlow).
    "general" is compatible with Caffe.
    The default is "valid".
  • pads : Integer(s), pad values to the input.
    Four integers mean [pad_top, pad_left, pad_bottom, pad_right].
    Two integers mean [pad_top, pad_left] where pad_bottom and pad_right are same as pad_top and pad_left respectively.
    One integer means that pad_top, pad_left, pad_bottom and pad_right have the same value.
    The default is 0.
  • has_relu : Boolean, whether the layer applies a ReLU function to the output.
    The default is false.
  • relu_max_value : Number, the max value of ReLU output.
    Valid only when has_relu is true and relu_max_value > 0.0f.
    The default is -1.0f, i.e. there is no max value.

Weights

  • kernel : 4D tensor [height, width, channels, filters], kernel (mandatory).

  • bias : 1D tensor [filters], bias (optinal).