Utility owned an abandoned coal pier on the Delaware River. They decided to not go forward with a half-million dollar security project they bid out to protect pier from riverboat pirates. We observed their bigger risk, which they had missed, was kids on dirtbikes entering from landside into a very hazardous area.
To really improve your security you have to learn to see what you don’t see. By this I don’t just mean “blind spots” or places where your cameras are not pointing, or where adding more would solve the problem. That would be way too obvious.
I mean you have to take a big step back and get the big picture of what your present security cameras and your overall security plan doesn’t do and can’t do.
To really understand this we will look at some technical things about video cameras, some psychological things about humans, and some statistical things about crime and losses at businesses, industry and institutions.
1. First of all, cameras don’t see anything.
To see you need a brain. Cameras just transmit an image. Understanding this distinction, obvious as it is, is crucial to vastly improving your security. How could something so obvious get overlooked by buyers of security systems? It is simple. A salesperson from a security company stands in your parking lot waving his or her arm in the air. You sit in your office looking at a monitor and say, “I can see them clear as day!” So you feel that if any thieves or wrong-doers are there, certainly your camera can see them.
But remember: Your camera didn’t see them — you did! And you happened to be looking at your monitor right at the moment they were standing there. That will not be the situation in the real world when a thief shows up. So, most people say that at least they will be able to go to the video recorder the next day when they have discovered the theft or arson or whatever happened, and see who did it. Really?
2. Did the salesperson demonstrating the camera stand 80’ away?
Let’s assume your outdoor camera is set to an angle of view of 60 degrees and is looking out over a parking lot that is 300’ deep. When you bought the camera, did the salesperson demonstrating it stand 300’ from the camera or was it 80’ away? Hmmm. In broad sunlight at 80’ you probably would not be able to forensically identify the person. That is, you couldn’t tell if it was Sally or Barbara. At 300’ away, with the subject standing still in broad sunlight, you might or might not be able to tell that it is a person, not a signpost. Why is it that you thought, during the sales demo, that you could see the person 80’ away clear as day?
Part of the reason is that human memory fills in all sorts of gaps in perception. Psychologists have proven that memory makes things up, to fulfill what seems like a logical narrative. You were talking to the salesman in your office. Then he walked out into your parking lot and stood 80’ away, facing the camera. Your mind filled-in that it was the same person and you could see his face. But if you check the videotape, you may find to your surprise that you could see no such detail.
3. Now for the real world.
But, worse yet, the above is based on ideal conditions. If the person at 80’ away — forget about the 300’ — was walking and it was nighttime, even in your illuminated parking lot, they would not gather enough light. On most video cameras they will only look like a blur or a ghost or a wisp of air. There may be rain, snow or glare, making it much worse. They probably will not be facing the camera like the sale demo person did. They may be wearing a ski mask or scarf over their face, but that would only be because they didn’t understand the limitations of most video cameras or they wouldn’t bother with the mask.
4. What is the answer?
The answer is artificial intelligence. “A.I.” for short. You can hook your existing video surveillance cameras up to an A.I. brain. If you don’t have cameras, you can get ones that already are connected to A.I. brains. Machine-vision is complex software that mimics how human vision works. A video camera, so enabled, also has much greater visual acuity or perception than humans do. I know this all sounds very Star-Treky but it has existed for about fifteen years. The computer brain will be able to reliably determine that it is a human intruder, off 300’ in the distance on your 60 degree angle-of-view camera. Even with rain, fog, or low levels of illumination, and even with the subject walking, not standing still gathering light.
Within seconds the A.I. sends an alert to you or to your security staff or to a video security real time monitoring central station. They don’t have to be trying to stare at a wall of video monitors displaying three hundred cameras! It alerts them and it highlights the intruder on the correct camera. Then they can talk down over an outdoor public address system, talking directly to the intruder. “This is Security. You are being video recorded. You have violated a secure area. If you don’t leave immediately the police are being called.” Not a canned recording, but a real person in real time. So we are not concerned with the forensic identification after a crime. We prefer to deter the crime before it happens. That is the big picture of the dramatic change in security that is now possible.
The statistical big picture is that losses from workers comp claims and public liability claims exceeds the cost of crimes. The right A.I. video security systems be equally effective in preventing those type of losses. Traditional video did little or nothing for that. Also, employee theft costs as much as outsider theft. Again, A.I. can deter that.