How to put face alignment into deepstream pipeline?

Face alignment is some image translation operations, like rotate the face to a target angle. We can implement them using OpenCV packages (like cv2.warpAffine) or other packages.
My question here is that, how can we put such operations into deepstream pipeline, and make the pipeline to be: face detect → landmark detect → face alignment → face recognition.

Is it possible? If yes, can you provide the resources/procedure/links?

Tao provides FaceDect directly. See FaceDetect — TAO Toolkit 3.0 documentation and https://ngc.nvidia.com/catalog/models/nvidia:tao:facenet
The model described in this card detects one or more faces in the given image / video.

Hi Morganh,

Thank you for your reply.
Maybe I didn’t describe the problem clearly. My question is not how to use face detection. My question is how to, for example, put some OpenCV algorithms into the Deepstream pipeline. Like the pipeline example I used: face detect → landmark detect → face alignment → face classification. In this pipeline, face alignment is some operations/algorithm from OpenCV. And all the others are Deeplearning models and can be supported by the Deepstream. So how can I put such OpenCV algorithm (non-deeplearning model, in fact it’s just some matrix transformation operations) into the Deepstream?

Could you create a topic in deepstream forum? It is a question for deepstream.