2019. 1. 24. 22:09ㆍ[정리] 직무별 개념 정리/딥러닝
def getConv1DOutputSize(inputSize, kernelSize, stride, padding):
output = int((inputSize + 2 * padding - kernelSize) / stride + 1)
return output
def getConv2DOutputSize(inputSize, kernelSize, stride, padding):
output = [getConv1DOutputSize(inputSize[0], kernelSize, stride, padding),
getConv1DOutputSize(inputSize[1], kernelSize, stride, padding)]
return output
def getConvT1DOutputSize(inputSize, kernelSize, stride, padding):
output = int((inputSize - 1) * stride + kernelSize - 2 * padding)
return output
def getConvT2DOutputSize(inputSize, kernelSize, stride, padding):
output = [getConvT1DOutputSize(inputSize[0], kernelSize, stride, padding),
getConvT1DOutputSize(inputSize[1], kernelSize, stride, padding)]
return output
print("Conv")
input = [64, 64]
print(input)
input = getConv2DOutputSize(input, 4, 2, 1)
print(input)
input = getConv2DOutputSize(input, 4, 2, 1)
print(input)
input = getConv2DOutputSize(input, 4, 2, 1)
print(input)
input = getConv2DOutputSize(input, 4, 2, 1)
print(input)
print("Conv Transpose")
input = [1, 1]
input = getConvT2DOutputSize(input, 4, 2, 0)
print(input)
input = getConvT2DOutputSize(input, 4, 2, 1)
print(input)
input = getConvT2DOutputSize(input, 4, 2, 1)
print(input)
input = getConvT2DOutputSize(input, 4, 2, 1)
print(input)
input = getConvT2DOutputSize(input, 4, 2, 1)
print(input)
'[정리] 직무별 개념 정리 > 딥러닝' 카테고리의 다른 글
| KLDivergence (0) | 2019.01.24 |
|---|---|
| pyTorch - Mnist VAE (0) | 2019.01.24 |
| Tensorflow - Mnist CNN 분류 (0) | 2019.01.24 |