Kitti formats only export projected objects

[jotamotk]

Hi, I can't find Lidar Basic feature in BasicAI Cloud. Could you explain how to use it, pls. I want to annotate Lidar and front camera data and export in kitti format, but with 'Lidar Fusion" feature, only projected object are exported. Thank you for your help, Regards Yvan

Originally posted by @e1eustac in https://github.com/xtreme1-io/xtreme1/issues/164#issuecomment-1968410760

[jotamotk]

Hi Yvan, We have merged LiDAR Basic datatype into LiDAR Fusion in our cloud version and have yet to migrate to Xtreme1.

If you have previous LIDAR Basic zip files, you can directly upload them to a LiDAR fusion dataset. See sample datasets here

We only support converting 3D results with the 2D results for Kitti formats. However, we can add a script for you; please provide the following information:

1. In which version are you trying to export, cloud or Xtreme1?
2. Please provide your expected Kitti-like format details

Regards, James

[e1eustac]

Hi James, Thank you for your answer. I will soon test the fusion basic feature in cloud platform.

1. I have tried to export with cloud and Xtreme1. With cloud platform, only projected objects are exported. With Xtreme1 and export script (xtreme1.script_ctl form xtrem1-sdk v0.6), kitti directory is empty. Example: python -m xtreme1.script_ctl --mode export --src "./LiDAR.zip" --dst "./exportKitti" --fmt kitti LiDAR.zip

2. Kitti format: sailing-yacht 0 0 0 0 0 0 0 11.5 3.0 10.5 -13.1977033 9.76468154 68.57108511 0.27449926	Values	Name	Description
   1	class	Describes the type of object: 'sailing-yatch', 'motor-boat', 'vessel', 'buoy' or 'DontCare'
   1	truncated (unused)	Float from 0 (non-truncated) to 1 (truncated), where truncated refers to the object leaving image boundaries
   1	occluded (unused)	Integer (0,1,2,3) indicating occlusion state: 0 = fully visible, 1 = partly occluded 2 = largely occluded 3 = unknown
   1	alpha (unused)	Observation angle of object, ranging [-pi..pi]
   4	bbox (unused)	2D bounding box of object in the image (0-based index): contains left, top, right, bottom pixel coordinates.
   3	dimensions	3D object dimensions: height, width, length (in meters)
   3	location	3D object location x,y,z in camera coordinates (in meters)
   1	rotation_y	Rotation ry around Y-axis in camera coordinates [-pi..pi]
   1	score	Only for results: Float, indicating confidence in detection, needed for p/r curves, higher is better.

Kind Regards, Yvan

[jotamotk]

Hi Yvan, We are working on a more universal Kitti format compatible with the scenario that can export results without 2D projection. We will close this issue when it is done. We will keep you posted!

Best, James

[jotamotk]

Hi Yvan, We have added a KITTI-like export script to our cloud version. You can find it by selecting the data format for export. For Xtreme1 you can check our sdk

Best, James

[e1eustac]

Hi James, Thank you for your efficient work. It's seem to be ok. Regards Yvan

[jotamotk]

Sure