Y
Hacker News
new
|
ask
|
show
|
jobs
by
rsp1984
3271 days ago
My guess is they estimate one or several dominant scene planes from the sparse triangulated feature points and get scale through incorporating accelerometer measurements.