Sorry in advance I tend to get off topic at times and then ramble. So this might be missing the entire point of the post, however rather than relying on your human eye to identify the individual’s vehicle by its license plate (Which is a Machine Learning/Artificial Intelligence Skill) The Wyze Cam is a Linux Kernel based module and should potentially be capable of running openCV (openComputerVision) which would offer your camera to utilize its native ability to fully operate the lens for comparing similarities such vehicle recognition, alterations from the stock image it used to identify the vehicle versus what the lens captured which would include parking permits on windshields, handicap parking permits, vehicle occupants, damage to the vehicle. etc. This would give you the ability to object track as well so if per say there is a main road through your neighborhood and the person would have to use that road to come into and out of your neighborhood, it would be able to identify all that into a log file to compare to at a later time. Human eyes are also UV filtered and IR blind and machine lenses turn purple when they get blinded with IR light and have to filter that light to continue operation.
I’m not sure if your municipality or local government accepts identification of individuals based on forensic data compiled by a computer. That has been something that has been challenged and currently it is no longer viable where I live in Ohio, the courts ruled that a Human Police Officer must be present to formally issue the citation, otherwise it was invalid no matter how much evidence there was proving the crime.