日本研發出智能軟件 不等下手就能識別小偷

A Japanese tech startup claims to have developed an artificial intelligence software that analyzes surveillance camera footage in order to detect suspicious behavior and prevent shoplifting before it actually happens.

日本一家科技初創企業聲稱研發出了一款人工智能軟件，能夠通過分析監控攝像頭拍下的視頻畫面識別出可疑行爲，在小偷下手前就加以阻止。

In 2002, when the movie Minority Report, starring Tom Cruise, came out, the concept of preventing a future crime before the perpetrator did anything remotely suspicious was nothing more than a cool sci-fi idea, but thanks to the rapid advancement of artificial intelligence, we already have an early version of the technology displayed in that movie.

2002年，湯姆·克魯斯主演的電影《少數派報告》上映，當時預知犯罪不過是個科幻概念，但隨着人工智能的迅猛發展，如今電影中展現的技術在現實中已經初見雛形。

VaakEye, an AI software developed by Japanese startup Vaak, can allegedly detect shoplifting before it occurs by analyzing surveillance camera footage for discrete suspicious movements and behaviors of people caught in the camera’s field of view. If the algorithm decides that there is a high-enough probability of a person shoplifting, it sends an alert via a smartphone app so the crime can be prevented.

日本初創企業Vaak研發的軟件VaakEye據稱可以通過分析攝像頭視頻中不連續的可疑動作和行爲來識別盜竊行爲。如果算法認定某個人偷竊的可能性足夠高，就會通過智能手機應用發送警報以阻止犯罪。

VaakEye first made news headlines back in December 2018 when it picked up on previously undetected shoplifting activity at a convenience store in Yokohama. The suspect, an 80-year-old man, was apprehended a few days later after stealing a hat. Today, Vaak is currently testing its software in a few dozen stores in the Tokyo area, but plans to have it installed in 100,000 stores across Japan in the next three years.

VaakEye首次登上新聞頭條是在2018年12月，當時該軟件在橫濱的一家便利店發現了此前未被察覺到的偷竊行爲。嫌疑人是一名80歲老人，幾天後因偷了一頂帽子被捕。如今，Vaak公司正在東京地區的幾十家商店測試這一軟件，並計劃未來三年內在日本各地的10萬家商店安裝該軟件。

"I thought then, ‘Ah, at last!” Vaak founder Ryo Tanaka recalled about the first time VaakEye helped detect shoplifting intentions. “We took an important step closer to a society where crime can be prevented with AI.”

Vaak公司的創始人田中亮回憶起VaakEye軟件第一次幫助識別偷竊意圖時說道：“我那時候想，終於做到了！我們距離用人工智能預防犯罪的社會又近了一大步。”

VaakEye is based on a complex algorithm that harnesses deep learning from more than 100,000 hours of surveillance footage and analyzes over 100 different factors, from a person’s face, clothes, movement, overall behavior, to macro-data like crime incidence in the area the shop is located in, or current weather conditions.

VaakEye軟件基於一個複雜算法，該算法對超10萬個小時的監控視頻進行了深度學習，分析了100多種不同因素，從一個人的臉、服裝、動作、整體行爲，到商店所在地區的犯罪率或當下天氣條件等宏觀數據。

This new shoplifting-prevention software has reportedly helped detect at least 7 shoplifters in tests performed as of December 12th last year, and is credited with reducing shoplifting cases by 75%, according to IT Media.

據IT Media媒體報道，截至去年12月12日的測試期間，這款新的防偷盜軟件已經幫助識別出至少7名小偷，並將盜竊案數量降低了75%。

However, VaakEye is not intended to help police catch shoplifters after they’ve committed the crime, but rather help business owners prevent the crime from being committed. Theoretically, shop staff can approach suspicious subjects after receiving an alert from the software and ask if they need some help. This should discourage suspected shoplifters from going ahead with their plans.

但是，VaakEye軟件的目的不是爲了在小偷盜竊後幫助警察抓住罪犯，而是幫助店主防止偷盜發生。從理論上來講，店主可以在軟件上收到警報後接近可疑人員並詢問他們是否需要幫助。這樣應該可以阻止嫌疑犯繼續實施犯罪計劃。

While Vaak is currently focusing on shoplifting prevention, they also have plans to enhance the software’s functionality. 30-year-old Tanaka told Bloomberg that it could also be used to analyze the way consumers interact with various products, thus helping businesses better display their goods, and also in public spaces and train platforms to detect suspicious behavior and prevent suicide attempts.

儘管Vaak公司目前專注於預防盜竊，但他們也計劃提高該軟件的功能性。現年30歲的田中亮告訴彭博社說，該軟件還可以用來分析消費者與各種產品的互動方式，從而幫助企業更好地展示他們的商品，也可以用在公共場所和火車站臺，以察覺可疑行爲，防止有人企圖自殺。

"We’re still just discovering all the market potential,” Tanaka said. “We want to keep expanding the scope of the company.”

田中亮說：“我們還在探索該軟件的所有市場潛力，希望能不斷擴大公司的業務範圍。”