Enable Azure Media Services streaming an analytics through IoT Edge.
Live media streams requires usually a good bandwith. By running the Azure Media Services, through the IoT Edge, it should be possible to employ ML, and vision analytics on the edge before streaming to cloud.
Micah Lewis commented
Take a look at the Live Video Analytics from AMS
I think there are plenty scenarios where this could be useful. Particularly when you need near real-time response, such as for process control etc.
There are also scenarios where you only want to stream media upon certain events e.g security camera events etc. in order to save bandwidth where resources are limited.
IFXE - Interesting ideas, I am curious what would you do with the resulting ML insights - perhaps try to auto-avoid "bad" routes, or maybe get a feedback loop via the ML to auto-reduce your bandwidth if that would improve the session, or maybe auto-report the "bad" route to your ISP?
Re: trying to achieve better live streaming results with AMS you might consider using the open source SRT technology (www.srtalliance.org).
Ping me directly if you'd like some more info on live encoding (with and without SRT) via AMS (zip @ haivision.com).