A groundbreaking AI technology has been unveiled to tackle the widespread problem of theft in retail environments. Known as Vaakeye, crafted by the Japanese startup Vaak, this system utilizes artificial intelligence to identify potential shoplifters by observing their body language, promptly informing store staff via an application to prevent criminal activity.
By analyzing footage from surveillance cameras, the system distinguishes between lawful customers and suspicious individuals engaging in behaviors such as nervousness and agitation. CEO of Vaak, Ryo Tanaka, elaborated that the AI system detects dubious conduct by examining various aspects of shoppers’ movements, such as their posture, gestures, attire choices, and facial expressions.
In a notable incident in 2018, the software assisted in apprehending a thief at a convenient store in Japan, signifying a significant achievement for Tanaka. He firmly believes that AI can be instrumental in preventing criminal activities, given that global retailers reportedly suffered around $34 billion in revenue losses due to theft in 2017, as reported by Bloomberg Quint.
The company, based in Tokyo, has commenced the sale of its theft-detection system and aims to have it installed in 100,000 stores throughout Japan within the next three years. Retail expert Thomas O’Connor from Gartner highlighted that numerous retailers in the US and UK already employ CCTV cameras in their establishments, and integrating them with a sophisticated tool like Vaakeye can notably enhance the efficiency and efficacy of theft prevention.
Nonetheless, a joint study by MIT and Stanford University pointed out the potential bias exhibited by commercial AI systems, including discrimination based on race and gender, influenced by the data used to train these systems. This issue has ignited concerns, exemplified by Amazon’s facial recognition software facing backlash for displaying biased outcomes. Despite this, Tanaka underlined that his company is just beginning to realize the lucrative prospects of their software.
Image Source: Andrey_Popov / Shutterstock
