Here’s my latest comic: https://inhumaneresourcescomic.com/#/panel/650.
So here’s a clearer version of what Finch is saying that doesn’t fit in the panel.
If you have a picture of someone from a driver’s license or social media post, and you try to find the person in Julius’ new analysis, it’ll succeed 76% of the time.
If you have video in the pre-Julius system and you try to identify that person with Julius’ new analysis, it’ll succeed 91% of the time.
Once someone is identified in Julius’ new analysis (using either of these methods or having someone go through the video, point and say, “Hey, that’s Steve”) it can keep identifying that person for future appearances more than 99.9998% of the time.