I recall - as a parent - using various books to help my kids learn about things that don't match. It's a cognitive exercise that encourages pattern recognition which will be useful every day of their lives.
The same approach is helpful for identifying ways to create safer environments. And the Internet of Recognition™ can provide us with some useful tools for creating safer environments in all aspects of society.
IBM's Safer Planet solutions and services is the inspiration for this article, but my thoughts are uniquely driven by practical ways that machine vision and resulting imagery analytics aided by deep learning can create greater safety in small but effective ways.
Things that DON'T Belong
Imagine a video camera strategically placed behind the counter at a KFC restaurant. It's used to monitor workers and their activities, not the guests. Further, imagine that each employee's face has been captured and the system has been configured to recognize everyone who works at this KFC location.
With video analytics and real-time processing, we can constantly monitor the video feed looking for people who have no business being behind the counter at this restaurant. In the blink of an eye, we can alert managers, security, or even local law enforcement that something out of the ordinary has occurred at this location.
Certainly, delivery and other maintenance personnel access the restaurant from time-to-time, but entry is generally through a secure back entrance. The front counter is where criminals gain access.
This simple approach can be applied to dozens of use cases by simply identifying people and things that don't fit a specific pattern and for which grave concerns may exist. Here are some others.
- An unattended bag in an airport.
- People walking on a frozen pond.
- Three people approaching one person in a park at night.
- A vehicle driving the wrong way on a one-way street.
- A worker standing on a maintenance walkway near a conveyor system that is about to start up.
- Liquid on the floor of a retail store.
- A vehicle driving without lights at night.
- Smoke emerging from a dumpster.
- A person standing next to a bus driver for a prolonged period of time.
- A child improperly restrained on an amusement park ride.
- Debris or people on a railroad track.
- People walking outside the yellow lines of an assembly line manufacturing floor.
- A vehicle parked in a fire access zone.
- A vehicle passing through a crosswalk when the traffic light is red.
Things that DO Belong
We can apply the same practical measures using imagery analytics to recognize situations where safety is compromised when things are missing.
- People without hard hats at construction sites.
- Missing fire extinguishers in a machine shop.
- DOT workers without neon vests.
- Construction and maintenance workers at heights without tethers.
- People without eye protection in proximity to dangerous chemicals.
- Health workers without bacterial masks in critical care areas.
Safe Processes and Patterns
There are likely to be hundreds of processes that, if not performed consistently, result in compromised safety. Machine vision and video analytics is now able to recognize unmistakable flaws in processes performed by other machines or humans and report deviations from expected and safe patterns.
Consider something as simple as healthcare workers washing their hands before approaching or touching patients.
Hospitals, of course, require this procedure and spend millions trying to enforce such policies. Insurance rates are measurably impacted by failure to adopt strict enforcement of hand-washing protocols. Super-bugs and other diseases are spread predominantly through hands in hospitals. But even with the most advanced RF, proximity, and motion-sensing devices, the best they can do is capture analytics that approximate actual practices.
Imagery analytics can detect when workers wash and precisely who they are. Using real-time notifications, workers can be alerted when they approach a patient without first washing their hands.
Such a system will ensure compliance proactively and likely result in vastly improved analytics.
Other process-oriented use cases for improved public safety can be easily imagined.
- Clearing firearms before surrendering them at court house check points.
- Visually accounting for dangerous chemical supplies, e.g., recognize who took them and when they were returned.
- Tracking inspection processes and their completeness.
- Commercial vehicle inspection before leaving the lot, e.g., video analytics can detect broken running and tail lights.
The Future is Bright
The Internet of Recognition™ will significantly improve public safety, but practical use cases will advance in subtle ways, much like artificial intelligence has seeped into every aspect of our lives without us realizing it.
Add paragraph text here.