[Study Notes] Week1_Convolutional Neural Networks_Strided Convolutions

1. Strided Convolutions

    1) After each convolution, the convolution kernel moves s = stride pixels

        

    2) The size of the image after convolution is: [(n + 2p - f) / s] + 1 + 1 *  [(n + 2p - f) / s] + 1

        Note: if [(n + 2p - f) / s] + 1  is not an integer then it is rounded to the left

        

2. Summary: Convolution Dimension

    

    You don't have to choose to make the result an integer, just round it to the left.