I find that the typical missing link to understand this response is trust in the police. If you trust them to do their job (and trust that their job is actually to protect the community, not attack it), you also typically trust in them to properly use the tools at their disposal. In that case, you typically think about the benefits of those tools and how they can let police do their job _better_ instead of how they could be abused or used for wrong.
If people don't trust the police (or think that their job/role is a negative one), then they probably tend to focus more on the potential downsides or bad possibilities that could come with new or improved tools (of which there are many with facial recognition). If you don't like what the police are already doing, it's easy to assume giving them the tools to do it better would be a Very Bad Bad thing.
You definitely don't have to accept heavy-handed police-driven violence (which I don't) to understand that it's important for the police to be able to do their job better. Ironically, I think that physically unobtrusive information-based solutions (like FR) that let officers do their job with more information and less unpredictability would result in _less_ physical violence.