[ 深度学习 ] 基础知识丨2. 卷积核；步长；padding的计算方法

完整版Hout=⌊Hin+2×padding[0]−dilation[0]×(kernel_size[0]−1)−1stride[0]+1⌋H_{out} = \left\lfloor\frac{H_{in}+ 2 \times \text{padding}[0] - \text{dilation}[0] \times (\text{kernel\_size}[0] - 1) - 1}{\text

小小的行者

11678人浏览 · 2021-06-05 20:53:29

小小的行者 · 2021-06-05 20:53:29 发布

完整版

$kernel_size [ 0 ] − 1 ) − 1 stride [ 0 ] + 1 ⌋ H_{out} = \left\lfloor\frac{H_{in} + 2 \times \text{padding}[0] - \text{dilation}[0] \times (\text{kernel\_size}[0] - 1) - 1}{\text{stride}[0]} + 1\right\rfloor$

$kernel_size [ 1 ] − 1 ) − 1 stride [ 1 ] + 1 ⌋ W_{out} = \left\lfloor\frac{W_{in} + 2 \times \text{padding}[1] - \text{dilation}[1] \times (\text{kernel\_size}[1] - 1) - 1}{\text{stride}[1]} + 1\right\rfloor$

来自：Pytorch-Conv2d

$kernel_size [ 1 ] − 1 ) − 1 ] [\text{dilation}[1] \times (\text{kernel\_size}[1] - 1) - 1]$ 是加了dilation之后的kernel size。

简易版

$kernel_size [ 0 ] stride [ 0 ] + 1 ⌋ H_{out} = \left\lfloor\frac{H_{in} + 2 \times \text{padding}[0] - \text{kernel\_size}[0]}{\text{stride}[0]} + 1\right\rfloor$

$kernel_size [ 1 ] stride [ 1 ] + 1 ⌋ W_{out} = \left\lfloor\frac{W_{in} + 2 \times \text{padding}[1] - \text{kernel\_size}[1] }{\text{stride}[1]} + 1\right\rfloor$