Convolution

class dragon.vm.caffe.layers.Convolution(layer_param)[source]

Apply the n-dimension convolution.

The spatial output dimension is computed as:

\[\begin{cases} \text{DK}_{size} = dilation * (\text{K}_{size} - 1) + 1 \\ \text{Dim}_{out} = (\text{Dim}_{in} + 2 * pad - \text{DK}_{size}) / stride + 1 \end{cases} \]

Examples:

layer {
    type: "Convolution"
    bottom: "input"
    top: "conv1"
    convolution_param {
        num_output: 32
        bias_term: true
        kernel_size: 3
        pad: 1
        stride: 1
        dilation: 1
        group: 1
        weight_filler {
            type: "xavier"
        }
        bias_filler {
            type: "constant"
            value: 0
        }
    }
}