2

For example let's say that this is the input image:

Input image

This is the original ground truth downloaded:

GT

But in all the papers they show something like this:

heat map

How can I obtain the last one?

aca06
  • 45
  • 1
  • 5

1 Answers1

1

In order to obtain a dense depth map, you need to run a depth inpainting/depth completion method on the Lidar data, which is the ground truth data you downloaded.

I suggest you to take a look at the following:

https://github.com/balcilar/DenseDepthMap

https://gist.github.com/ialhashim/be6235489a9c43c6d240e8331836586a

Also the works under the KITTI Depth Completion Benchmark could be helpful:

http://www.cvlibs.net/datasets/kitti/eval_depth.php?benchmark=depth_completion

doldnh
  • 312
  • 3
  • 12
  • I'd like to point you to my codebase as well https://github.com/soulslicer/kitti_depthmap – raaj Mar 07 '20 at 19:20
  • The issue with balcilar's code is that it doesnt handle parallax errors well. Also, that code is written in Matlab, and we all know what a miserable language that is. My code is in Python/C++ hence is fast, and easy to integrate into your ML pipeline – raaj Mar 07 '20 at 19:20
  • @raaj, what is the `utils_lib` package that you're importing in your code in `testing.py`? – Nagabhushan S N Mar 19 '21 at 19:07