The Euclidean distance degree formula for multiview camera resectioning

Let $X_{\bar{\mathbf{q}}, n}$ be the affine variety of image configurations arising from a camera and world points $\bar{\mathbf{q}} \in (\mathbf P^3)^n$. For generic data, let $\operatorname{ED}(X_{\bar{\mathbf{q}}, n})$ denote the number of critical points of the squared Euclidean loss restricted to the smooth locus of $X_{\bar{\mathbf{q}}, n}$. Euclidean distance degree conjecture. For all $n\ge 6$ and generic $\bar{\mathbf{q}} \in (\mathbf P^3)^n$,

$$\operatorname{ED}(X_{\bar{\mathbf{q}}, n}) = (80/3)n^3 - 368n^2 + (5068/3)n - 2580.$$

This formula predicts the algebraic complexity of globally optimizing the geometric-error objective in camera resectioning; the supplied source does not indicate whether the claim has been proved or disproved.