feature map size 계산

---

from tensorflow.keras.layers import Input, Conv2D
from tensorflow.keras.models import Model

input_tensor = Input(shape = (5, 5, 1))
x = Conv2D(filters = 1, kernel_size = (3, 3), strides = 1)(input_tensor)
print("x's shape : {}".format(x.shape))

---

from tensorflow.keras.layers import Input, Conv2D, ZeroPadding2D
from tensorflow.keras.models import Model

input_tensor = Input(shape = (6, 6, 1))
padded_input = ZeroPadding2D(padding = ((1, 0), (1, 0)))(input_tensor)
x = Conv2D(filters = 1, kernel_size = (3, 3), strides = 2)(padded_input)

print("x's shape : {}".format(x.shape))

padding을 kernel의 위, 왼쪽에만 해준 상황이다. 이러한 경우는 공식에서의 S가 0.5가 된다.