New Python example for RealSense SDK 2.0 - box measuring with multiple cameras

[MartyG]

Hi everyone,

The Python wrapper of RealSense SDK 2.0 has a new Python example for dimensioning boxes with multiple cameras.

https://github.com/IntelRealSense/librealsense/blob/master/wrappers/python/examples/box_dimensioner_multicam/box_dimensioner_multicam_demo.py librealsense/box_dimensioner_multicam_demo.py at master · IntelRealSense/librealsense · GitHub

[PMalt3]

I noticed that the above example makes good measurements only at a certain position of the camera.

Good measurements are obtained when:

1) the direction of the camera relative to the vertical should be significant, close to 45 degrees;

2) the distance from the camera to the object should be approximately 30 centimeters.

If you do not fulfill paragraph 1, then the height is much less than the actual values ​​(two times)

If paragraph 2 is not fulfilled, all sizes are very different from the real ones, and the point cloud is unstable.

I just can’t understand what is the reason for this code work. Can someone help me?

The screenshot shows the very unsuccessful result (and not the worst)

[MartyG]

You did not mention how many cameras you are using. This program is designed for multiple cameras and does automatic calibration of those cameras when the program starts. So if only one camera is used, the measurements may not be as good as a multiple camera setup.

How close you can get the camera to the object before the image starts breaking up will depend on the camera model. The minimum default distance of D415 is 0.3 m (30 cm) while the minimum of D435 is 0.2 m (20 cm).

Also, I recommend centering your box on the checkerboard surface instead of having the box at the edge of it.

If your checkerboard is not 6x9 in size then you should adjust the values in the program.

https://github.com/IntelRealSense/librealsense/issues/3540

If you would be interested in a pre-made RealSense box measuring solution, the LIPSMetric Parcel Kiosk product recently became available.

https://github.com/IntelRealSense/librealsense/issues/5673