General forest type can probably be determined quite well and I imagine that ecologists have been estimating tree mass of various types of forest for quite some time?
Recent developments with laser scanning using drones is likely to improve accuracy too.
The whole point of the article is that they've (hopefully) improved the accuracy by using recent developments and combining different types of counts, and that method dramatically increases the count.
I'm sure we can count trees from pictures reasonably well, but we can't count tree mass that's entirely obscured by other objects.
To count trees accurately from pictures would require a vast quantity of footage taken from under the canopy.