Hacker News new | ask | show | jobs
by skywal_l 447 days ago
Right there in the abstract:

"Our approach works by predicting a dense motion flow field and a monocular depth map directly from a single motion-blurred image"