In general, to apply segmentation to a camera, you start by creating a segmentation texture in the Resources panel. Then, you only need to apply that segmentation texture to a camera's Mask Texture field. The camera's output will then be masked by the segmentation texture.

Multiple segmentations involves doing the above multiple times. We created an example project to demonstrate this.

This project has two perspective cameras for 3D objects and two orthographic cameras for billboards. In each camera set, there is one for portrait background segmentation and one for portrait hair segmentation. If you're adding new objects to this project, make sure to add it underneath the correct camera type and then set its Layer (with object selected, top of the Inspector panel) to either segmentationPortrait or segmentationHair.