That was a question I had, too, because I imagine this isn’t currently possible.
I agree, and it made me think of this:
I wondered if a similar process might work this way:
- The user (doing continuous microSD recording to a large card with events tagged in the recording timeline) submits a query to AI Video Search.
- The Edge AI on the camera suggests candidate event recordings on local storage and submits these to Cloud AI asking for confirmation of a match.
- Cloud AI analyzes them and lets the app know which events it “thinks” match the user’s request.
- The app displays resulting thumbnails to the user to tap and review.
I don’t know if anything like that is technically feasible. I think what @carverofchoice suggested might be more realistic.
Yeah, definitely a good thing to ask about.