is there option to run my model on top of deepstream SDK


I’m currently working with deepstream sdk on jetson xavier. We have custom python developed and added few features like deepsort-yolo tracker, face recognitions. is there any possible or suggestion to run our code on top of the deepstream sdk.


Is it possible to implement your code with C++?
If not, you will need to compile the deepstream app into an executable and launch it via system call in python.

More, there is a feature tracker in the DeepStream SDK. It’s recommended to give it a try.
We also achieve face recognition with a detector+classifier before.


I am new to deepstream and wanted to make a custom pipeline where I want to add secondary inference on detected objects ROI. The secondary model gives out feature set(encoding) on the detected ROI of primary inference output. The secondary model is basically a face recognition model.
Can I use the available PLugins(nvinfer) to achieve the task?
If not how to proceed?