According to the Sensormatic Global Shrink Index, shoplifting cost the global retail industry about 100 billion dollars last year. That amounts to approximately 2 percent of revenue. While it doesn’t seem like much it could be crippling to chains that operate on razor-thin margins and mom and pop retail stores that don’t generate the revenue of their larger counterparts. One company is looking to use artificial intelligence and deep learning to spot shoplifters before they commit the actual crime.

Vaak Software Can Identify Potential Shoplifters

Tokyo based startup, Vaak has developed detection software that can help spot potential shoplifters. The AI analyzes real-time footage from security cameras looking for any behavior that may be suspicious. If the algorithm picks up something it will alert store staff via a smartphone app.

The algorithm analyzes more than 100 features such as facial expressions, body language, movement patterns, and clothing. Shoplifters often have telltale signs before they commit a crime like restlessness, oversized clothing, a flush face, or abnormal sweating. They also may be moving around or “casing” a certain part of a store, looking for security cameras and store personnel.

The algorithm will try to pick up on all of these indicators. If a staff member is alerted and approaches the person to see “if they need any help”, it may help thwart a possible shoplifting attempt.

Shoplifting Detection Software Catches Perp in Yokohama

According to the company’s website, the algorithm also takes into account things like crime rate in the area, weather conditions and time of year. For example, shoplifting rates increase over the holiday season.

The Vaak software actually helped catch a shoplifter in a convenience store in Yokohama last year. The company was actually just testing the software in the store and it ended up picking up on previously undetected shoplifting activity. The person was arrested a few days later.

Vaak Software To Enter Market This Month

The company has been testing the software in a few dozen stores in Tokyo. Vaak plans to start selling the shoplifting detection software this month and is hoping to be in 100,000 stores across Japan in three years. The software isn’t just for retail. Vaak hopes to expand use to public places like train stations to help detect suspicious behavior or people who may be suicidal.

Check out our articles on AI that attempts to spot angry customers and facial recognition helping to protect chimpanzees.