Deploying own dnn model on drive orin

Autonomous Vehicles DRIVE AGX Orin DRIVE AGX Orin General
1

Please provide the following info (tick the boxes after creating this topic):
Software Version
DRIVE OS 6.0.10.0
[1]DRIVE OS 6.0.8.1
DRIVE OS 6.0.6
DRIVE OS 6.0.5
DRIVE OS 6.0.4 (rev. 1)
DRIVE OS 6.0.4 SDK
other

Target Operating System
[1] Linux
QNX
other

Hardware Platform
DRIVE AGX Orin Developer Kit (940-63710-0010-300)
DRIVE AGX Orin Developer Kit (940-63710-0010-200)
DRIVE AGX Orin Developer Kit (940-63710-0010-100)
DRIVE AGX Orin Developer Kit (940-63710-0010-D00)
DRIVE AGX Orin Developer Kit (940-63710-0010-C00)
[1] DRIVE AGX Orin Developer Kit (not sure its number)
other

SDK Manager Version
2.1.0
[1] other

Host Machine Version
native Ubuntu Linux 20.04 Host installed with SDK Manager
native Ubuntu Linux 20.04 Host installed with DRIVE OS Docker Containers
native Ubuntu Linux 18.04 Host installed with DRIVE OS Docker Containers
other

Issue Description
Hi team, I want to build a model in pytorch and deploy it on Orin, what all are the steps I can follow?
My understanding is, I have take a pytorch model, convert it into ONNX and then using tensorRT optimization tool, I have to convert this ONNX to tensorRT. Correct me if I am wrong
Please help!

4

Dear @akshay.tupkar ,

Yes. You are correct.

6

Dear @SivaRamaKrishnaNV
but my concern is how can I deploy this model in sample dnn plugin? How can I feed the input to the model?

7

Dear @akshay.tupkar ,
May I know if you can generate TRT model using tensorRT_Optimization tool or it has non natively supported layers? Are you using any object detection model?
You need to replace the TRT model in the sample and also make changes in post processing as per model. Please see DriveWorks SDK Reference: DNN for reference.

8

Dear @SivaRamaKrishnaNV
I am using digit recognition model (MNIST) which is trained in pytorch.
And yes, I can generate TRT model using tensorRT_Optimization tool

9

Dear @akshay.tupkar ,
Please refer to DNN plugin and object detector sample as reference to load TRT model and perform inference. You may need to make few changes in preprocessing/post processing based on the model. Please file a new topic with model and your code in case you notice issue.

11

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.