My house is located on a corner lot in an extremely quiet subdivision and my street has an intersection roughly in-line with my front door. I have a soffit mounted camera located at one end of the house that captures a good portion of the length of the house, walkway, and front yard. There's a small portion of the street that is also in the camera's field of view. I'm using regions to exclude the street from the detection area and this very works well during the day. Night time detection is a different story. Headlights coming towards my house from the intersection regularly trigger my people detection rule. Luckily, there's so little night time traffic that this only occurs a couple of times per night. If I were on a busy street the false positives would be unbearable.
Looking at the image attachments included on the alert emails has me wondering how people detection works given that they look nothing remotely close to human.
What's the suggested work around for this?
