this issue seems to not be solved!
my main problem is i’m not able to use the ObjectDetector class with the following statement
model = ObjectDetector('ssd_mobilenet_v2_coco.engine')
i got an error
'NoneType' object has no attribute 'create_execution_context'`
I searched for days for a applicable solution.people have same issue like https://github.com/NVIDIA-AI-IOT/jetbot/issues/96
the reason for this error is Tensorrt version.
i used the sd card image jetbot_image_v0p3p2 with no problem but the camera feed hangs and stops updating the frame while jetbot_image_v0p4p0 works great with camera but produce the mentioned error.
Sorry that there are some dependencies between package and GPU driver.
So you cannot just downgrade the TensorRT package without re-flashing the system.
A possible workaround is to recompile the ssd_mobilenet_v2_coco model on the jetbot_image_v0p4p0 directly.
You can find the detail steps in this Github: https://github.com/AastaNV/TRT_object_detection
Hey @AastaLLL can you please guide me through downgrading tensorrt 7 to tensorrt 6. The cuda version is 10.2 libcudnn 7.6.5. The tensorrt 6 supports these versions. I have all these version with tensorrt 6 installed on my ubuntu desktop. I want to do the same for jetson nano. I don’t want to reinstall an older jetpack. The onnx parser in tensorrt7 is not working and i am having the same problem as above user. I need tensorrt 6.
As mentioned previously, there are some dependencies between package and GPU driver.
So you cannot just downgrade the TensorRT package without re-flashing the system.
You need to reinstall the older JetPack.