Generating trainings data with Isaac SIM for object detection training


i am currently training object detection following the detect_net example . When i run the provided sample scene under “packages/Nvidia/Samples/ObjectDetection/”, it always produces the same image twice. Is there a way that i can produce the image just once ?
I already played around with the “Frame Interval” parameter under the procedual-> RandomizerGroup. But the minimum i can set here is 1, and that changes the scene just every 2nd frame. If i set it higher, the scene changes less often. Is there a way that i can change the scene every frame?
Thanks in regards

Hi markus, there’s a bug in the RandomizerGroup.cs in the 2019.3 release version that causes this behavior. The fix is to modify packages/Nvidia/Core/Scripts/Runtime/Generation/RandomizerGroup.cs line 118

if ((randomizeOnUpdate && frameCounter > frameInterval) || Input.GetKeyDown(KeyCode.R))

Replace the > with >=.

This is fixed in the recent 2020.1 release which can be download from

Thanks for using Isaac. Please let me know if you have other questions.