Encrypt App for deployment

Please provide complete information as applicable to your setup.

• Hardware Platform GPU - GTX 2070
• DeepStream Version - 5.0.1
• TensorRT Version - 7.0.0.11
• NVIDIA GPU Driver Version (valid for GPU only) - 440
**• Issue Type - questions **

Hi

I have created a custom app in python using deepstream-app3 and deepstream-imagedata-multistream sample. Also created a custom model through nvidia toolkit in detectnet resnet10.
At the moment i am executing using python3 <script_name>.py
How can i convert the python3 file to deepstream-app so i can execute like deepstream-app -c <custom_app>.
Or is there a way to encrypt the app which can distributed for demo purpose.

Thanks

Hi @priyanshthakore ,
Sorry! A little confused.

“How can i convert the python3 file to deepstream-app so i can execute like deepstream-app -c <custom_app>.” ==> this does not encrypt the model.

If you want to encrypt your model, a simple wrok around solution is,

  1. add watermark on your model
  2. modify nvinfer source code to remove the watermark when loading the model, and build a new nvinfer plugin library

Hi
Thanks for the model work around solution.
But what about the python script ?

There are text file in /opt/nvidia/deepstream/deepstream-5.0/samples/configs/deepstream-app/ in this location that we execute by deepstream-app -c but there are no code of this app so how to convert our python script to something like this.

For example i have python deepstream app with people count with mask logic i need to demonstrate to someone but don’t want to send the source code i.e. python file to demo it .

Encrypt python script is not out of DeepStream support.
This is a common question for Linux, maybe you could search it on nethwork.

ok, i am aware of few such as pyamor to encrypt python scripts but was looking for the entire solution for deployment of scripts and models.
Will use docker to deploy.

Thanks for the help