PyCuda logic error on jetson nano with latest image about tensorrt_demos

[SaddamBInSyed]

@jkjung-avt
Also I am facing the issue similar to below

https://devtalk.nvidia.com/default/topic/1056268/tensorrt/tensorrt-do_inference-error/

Is there any particular reason for not implementing the yolo_async sample code?

I see only ssd_async code in the example.

[jkjung-avt]

A "trt_yolov3_async.py" is very easy to implement. You could just copy my "trt_ssd_async.py" and modify a few lines. I remember I did that, but saw very little improvement on FPS. So I did not put the code into my repository.

Regarding the problem you mentioned in the first post, it's likely because you did not initialize pycuda driver properly. Please read the documentation or reference my "trt_ssd_async.py" code.