4 ms·
Isn't a logical next step to extract the depth field? Possible?
Isn't a logical next step to extract the depth field? Possible?
Isn't the depth decoder part of the processing already?
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"