2-step Yolo Detection in Deepstream

Hi,

Does the Deepstream SDK have the capability to load a 2-detector model based on tinyYolo?

My use case involves plugging in to the surveillance system and identifying persons and a trait of that person, such as if he/she is wearing a facemask or not.
I tried the default object detection model, having the object classes - person, facemask, nomask detected at the same time, but what happens is that the model either detects person or facemask/nomask. I was kind of hoping it can identify both at the same time - person and facemask, person and no mask.

Thanks.