about the tda4vm-sk demo show?

zhang cheng

Other Parts Discussed in Thread: TDA4VM

hello,

when i run the demo, some mistakes occur like follows, but i can't find any mistakes, and this is normal in the past time.

it keep this state and don't have any reaction.

over 2 years ago

0 Takuma Fujiwara over 2 years ago

TI__Guru 50853 points

Hi Zhang,

Can the pipeline be ran with the environment variable GST_DEBUG=4, and collect the resulting logs? The command will be:

GST_DEBUG=4 ./app_edgeai.py ../configs/object_detection.yaml

Also, to help reproduce the issue on our side, could I ask the following:

Which SDK version is this using?
Which model is this using? And is it a custom model?
Has any modifications been done to the demo application?

Regards,

Takuma

0 zhang cheng over 2 years ago in reply to Takuma Fujiwara

Prodigy 115 points

Hello, thanks for your reply.

i will tell you about the 3 questions, first the SDK version is Processor SDK Linux for Edge AI 08.04.00, second, it is a custom model that is compiled on benchmark(TI) successfully, third, i just modify the inputs size when export the onnx model because i want to verify the speed of yolov5l on tda4vm whose image inputs are different.(from 640 to 960)

that's all, expect for your reply soon.

0 Takuma Fujiwara over 2 years ago in reply to zhang cheng

TI__Guru 50853 points

Hi Zhang,

Could you also post the logs from running the following?

GST_DEBUG=4 ./app_edgeai.py ../configs/object_detection.yaml

From the current information, my assumption is some changes in the model needs to be made (perhaps there is a parameter that was not configured). However, the additional logs will greatly help in narrowing down what the issue is.

Regards,

Takuma