Conv2D:
参数为 p a d d i n g = ′ v a l i d ′ padding='valid' padding=′valid′时, o u t p u t = ⌊ i n p u t − k e r n e l _ s i z e ⌋ / 2 + 1 output=\lfloor input-kernel\_size \rfloor/2+1 output=⌊input−kernel_size⌋/2+1 参数为 p a d d i n g = ′ s a m e ′ padding='same' padding=′same′时, o u t p u t _ s i z e = i n p u t _ s i z e output\_size=input\_size output_size=input_size
Conv2DTranspose:
参数为 p a d d i n g = ′ v a l i d ′ padding='valid' padding=′valid′时, o u t p u t = i n p u t × s t r i d e s output=input\times strides output=input×strides 参数为 p a d d i n g = ′ s a m e ′ padding='same' padding=′same′时, o u t p u t = ( i n p u t − 1 ) × s t r i d e s + k e r n e l _ s i z e output=(input-1)\times strides+kernel\_size output=(input−1)×strides+kernel_size