How do I convert depth image to point cloud?

[impet14]

---

Required Info
Camera Model	{ D400 }
Firmware Version	(Open RealSense Viewer --> Click info)
Operating System & Version	linux
Kernel Version (Linux Only)	(e.g. 4.14.13)
Platform	PC/Raspberry Pi/ NVIDIA Jetson / etc..
SDK Version	{ legacy / 2.. }
Language	{C/C#/labview/nodejs/opencv/pcl/python/unity }
Segment	{Robot/Smartphone/VR/AR/others }

How do I convert depth image to point cloud? Now, I am doing object detection from color then I get the region on the depth image. I would like to convert dept image to PC. Can anyone give me suggestion?
Thanks

[summer1216]

In examples of this library there's point cloud generation in OpenGL. In wrappers there's wrapper to convert realsense point could into pcl point cloud.

[MartyG-RealSense]

Have you seen the step by step tutorial on generating a point cloud? I believe that is what Summer1216 is referring to above in regard to 'Examples'.

https://github.com/IntelRealSense/librealsense/tree/master/examples/pointcloud

[impet14]

Do we have python example?

[MartyG-RealSense]

There is an example for saving point clouds as PLY files directly from Pyrealsense.

https://github.com/IntelRealSense/librealsense/blob/development/wrappers/python/examples/export_ply_example.py

[impet14]

How can we visualize point cloud using python?

[MartyG-RealSense]

I searched very hard but was unable to find a good, clear example of creating a point cloud in Python. One of the Intel support people on this forum such as Dorodnic or RealSense-Customer-Engineering may be able to answer this better, as they have access to resources that I do not.

The best I could find was a page in Japanese, where they apparently create a Python point cloud with a D415 by importing the Point Cloud Library (PCL).

https://qiita.com/mizumasa/items/c6ca2c73fa86c9a5a39c

[baptiste-mnh]

I'm also searching an efficient way to visualize in real time the generated pointcloud from the D400 cam from python wrapper.
It's possible with the open3D visualizer but way too slow for now. I recently add two functions in the development branch of librealsense SDK to make it works. : isl-org/Open3D#473 (comment)

You can have a look at python-pcl which is a python binding of the PCL lib but I did not manage to install it...

So I'm trying to find a way to create my own viewer using OpenGL and the python wrapper by getting the vertices and textures coordinate in numpy.
I'll start to apply the same
The exemple in C++ is a good start, I'll try to do the same process in python : https://github.com/IntelRealSense/librealsense/blob/master/examples/pointcloud/rs-pointcloud.cpp

If anyone have a clue about how to do it or a better way to visualise the point clouds It could be very usefull!

[dorodnic]

@baptiste-mnh thank you for the feedback, looking forward to your findings.
In parallel, I've been looking at point-cloud visualisation inside Jupyter notebooks using pyntcloud:

[MartyG-RealSense]

Thanks muchly for the point cloud tutorial, Dorodnic - I'll make sure the link gets promoted and passed on to those on the RealSense Support forum who have been asking about Python point clouds.

[RealSense-Customer-Engineering]

[Realsense Customer Engineering Team Comment]
@impet14
Still have any question related to this topic?