DrSleep / multi-task-refinenet

Multi-Task (Joint Segmentation / Depth / Surface Normas) Real-Time Light-Weight RefineNet

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

How to get Depth Coefficient (5000) as you used in your Paper

ravivaghasiya1998 opened this issue · comments

Hello
I was wondering that how did you calculate the depth_scale = 5000 and how did you calculate max and min depth.

I assume you are asking about the NYU dataset (for KITTI the situation is similar just the values are different).

I don't really remember where the 5000 scaling parameter is originally coming from, but the idea behind it is that dividing the values in the depth images by that scaling parameter will give you depth in meters.

The min/max depth in the notebooks are set purely for visualisation purposes, iirc the NYU dataset has the maximum depth of 10m since it is an indoor scenes dataset