I used a combination of the above.
Cams saw other cams, Optex PIRs, driveway sensors and tinkered with geophone sensors buried adjacent to the walk.
Old home in the 80's-90's I used optical beams and had many false positives mostly during bad weather and lightning...they did work fine for the driveway. (it was maybe 100 feet of driveway and I used one beam sensor next to the front of the house and one closer to the garage). It was a simple set up. Camera on the garage, on the side of the house and in the front of the house.
The Geophone sensors worked the best for traffic on an expressway about a mile away.
There was much more space though between the house and the end of the property.
No traffic on the street and night triggers (rarely) were coyotes and rabbits playing and teenagers walking around at 3 AM.
One night (happened for a few nights) there was a group of teens going driveway to driveway opening unlocked doors in automobiles left in driveways. (one entrance subdivision of 100 homes surrounded by a golf course - they walked across the golf course to get to the subdivision). They checked out one vehicle that was in my driveway. Daylight happened, one teen looked at the lights and directly at one camera (doesn't look like a camera) that night.
The PIRs triggered daylight after hours. Personally the daylight at night worked the best.
I was initially using text to speech alerts then went to chimes...one if by land two if by sea (1-north, 2-south, 3-east, 4-west).
Helping an automator peer a bit with perimeter detection. Mostly now because of an event like this...
This person was being chased by police and cut through the yards in the neighborhood.
Typically they enjoy their back yard around the time that this happened. The recorded event was a video trigger from ZM.
There are two cameras present facing the yard on opposite sides. They do not see each other today.