If you have $(x,y,z) \in \mathbb{R}^3$ and you relate it to $(x/z, y/z) \in \mathbb{R}^2$ then you have interpreted $(x,y,z)$ as one possible representation of the 2D vector $(x/z, y/z)$ in homogeneous coordinates. The special case of $z=0$ corresponds to "points at infinity" or directions. If you have $(x,y,z,w) \in \mathbb{R}^4$ and you relate it to $(x/w,y/w,z/w) \in \mathbb{R}^3$ then you have interpreted $(x,y,z,w)$ as one possible representation of the 3D vector $(x/w, y/w, z/w)$ in homogeneous coordinates. From this it becomes clear that to represent an $N$-dimensional point in homogeneous coordinates you need $N+1$ components.
To clarify what I mean by one possible representation, consider the following example:
$$(3,7) \equiv (3,7,1) \equiv (6,14,2),$$
where the second and third vectors are different representations of the same 2D vector in homogeneous coordinates. In general, if you have a 2D vector $(X,Y)$, then for any $\lambda\ne 0$, the vector $(\lambda X, \lambda Y, \lambda)$ is a representation of $(X,Y)$ in homogeneous coordinates. So you really get an equivalence class of points in $N+1$ dimensions corresponding to a specific point in $N$ dimensions. Notably this equivalence class forms a line through $(0,0,0)$ and $(X,Y,1)$ in the $N+1$ dimensional space.