In Andrew Ng's deep learning course on Coursera, how is a single scalar value obtained from a flattened image (feature vector)? First there is $w.T$ of shape $(1, n_X)$ which is multiplied by $X$ of shape $(n_X, 400)$, so by the laws of linear algebra, the remaining vector is of shape $(1, 400)$. This is then passed through Sigmoid to form vector A, and the shape remains at $(1, 400)$. How is this vector then converted to a binary scalar value ($0$ or $1$) for prediction $y$-hat?
$n_X$ is the number of feature, $400$ is the number of data.
Each of the entry of $A$ is the output of the sigmoid layer, it is between $0$ and $1$. We can then decide a threshold (typically $0.5$) such that if it is at least the threshold, we map it to $1$, otherwise, we map it to $0$.