I recently digitized some old home movies from my grandparents and looking through the film I realize I recognize almost no one. This got me thinking, I should go through this with my mother to help me identify family members as she's probably the last person who knows who anyone is.

Obviously I can just make a document with notes ex: 2:53 screen right, great uncle john"
But I'm hoping there's a more elegant solution in which I can keep info with the file and not some auxiliary document that dosent live in the same place.

I could just dump all that info into Metadata but that's maybe the lowest quality solution.
The xray function of Amazon prime video is the example of the best solution in a perfect world. (if you don't know, when you pause or bring up the context menu on prime video, there's a bar that identifies everyone one scene with a matching headshot)

I actually looked into it and there's an API that amazon uses for this called "RecognizeCelebrities" and uses images to train the program to recognize faces and populate it's menu. I'm not sure if it's open source but even if it was, it's over my level to use it practically I think.

The last option I can think of is using a video markup platform like frame.io but that's not really a long term solution.

Any ideas?