Increasingly, smart surveillance cameras are monitoring public places in search of suspicious cues, a high-tech version of “if you see something, say something.” By reviewing massive volumes of ordinary surveillance tape, algorithms can “learn” what type of behavior is typical right before a crime or terrorist attack is committed – like a person suddenly breaking into a run or abandoning a suitcase on a subway platform – and alert authorities.
Esther Hovers, a photographer, captures some examples of the seemingly deviant behavior that these cameras pick up in a photo exhibition called “False Positives.” These photographs, which Hovers took in Brussels, the de facto capital of Europe, are montages, partially natural and partially staged, which Hovers created by combining images from several minutes of video.
See if you can spot the deviant behavior in the photographs below (some photographs have more than one). Then scroll down for the answer. This picture contains an example of what Hovers calls "Anomaly #1":
Anomaly #1: Standing Still. The photo above shows one behavior smart surveillance cameras are taught to look for -- a person standing in a location pedestrians usually continue to move through.
Here's Anomaly #2:
Anomaly #2: Fast Movements. The photo above shows one example of a suspicious activity -- suddenly breaking into a run.
Here's Anomaly #3:
"Anomaly #3: Lonely Objects." Did you notice the suitcase in the photo above?
Here's Anomaly #4:
Anomaly #4: Placement on a Corner. According to security experts, standing on a corner can be a suspicious sign. Normally, pedestrians stand at the cross-walk, or a few feet up the curb in order to catch a taxi.
Here's Anomaly #5:
Anomaly #5: Clusters Breaking Apart. In the photo above, a group of people suddenly disperses into different directions.
Here's Anomaly #6:
Anomaly #6: Synchronized Movements. Suspicious behavior also includes a group of people moving in a way that seems unusually coordinated -- at the same tempo or in the same direction, for example.
Here's Anomaly #7:
Anomaly #7: Repeatedly Looking Back. Another suspicious sign is repeatedly looking over one's shoulder, as the photos above show.
And finally, Anomaly #8:
Anomaly #8: Deviant directions: The photos above show an unusual pattern in the way people are moving. "Context is very important, says Hovers. "It's an anomaly if everyone in a street is walking the same direction, and one person is walking the other way."
If some of these behaviors seem relatively innocuous, that is partly Hovers' point. While smart cameras offer big benefits in security, they also increase surveillance of behavior that is slightly unusual, but not criminal in any way. Like many new technologies, smart surveillance systems may contain worrying consequences for privacy and public freedom.
For one, these cameras currently make a lot of mistakes. Hovers says that nine out of 10 alerts that these systems issue today are “false positives” – what the industry calls false alarms.
Part of it is that algorithms are much worse than humans at recognizing context, says Hovers. For example, a smart surveillance system might alert authorities if foot traffic on a street reaches much higher volumes than normal – but a human would be able to figure out that the cause is a newly opened market, or a town festival.
Hovers says her project is more about future possibilities than the current state of security, since the vast majority of cameras in use today are not yet smart cameras. But Washington, Boston, Chicago, Amsterdam and other cities have begun testing out smart surveillance technology.
While everyone wants security, Hovers says she is concerned about the kind of judgments this system imposes on a society, and whether it would restrain some types of "abnormal" public expression -- of which art could be considered one.
“Not every type of deviant behavior is criminal behavior, and I’m happy about that, actually,” she says.
You might also like: