incorrect disparity generated from depth because of outliers in depth

EPFL-VILAB / omnidata

A Scalable Pipeline for Making Steerable Multi-Task Mid-Level Vision Datasets from 3D Scans [ICCV 2021]

Other

395 stars 49 forks source link

incorrect disparity generated from depth because of outliers in depth #28

Open Tord-Zhang opened 1 year ago

Tord-Zhang commented 1 year ago

Hi, I tried to convert the depth prediction to disparity. But I found that in most cases, there are some outliers in the depth predictions that make the disparity incorrect. Is there any postprocessing that can fix this?

outliers