Artificial intelligence combined with traditional retail has produced a chemical reaction and reconstructed people, goods and fields. The rise of new retail terminals such as unmanned stores has enriched and supplemented commercial formats, and promoted the development of commerce to be smarter, more convenient, and more user-friendly.
On January 18th, China (Shanghai) Smart Retail Summit Forum was successfully held at Shanghai Hongqiao International Convention Center.
Gu Jiahe, former deputy director of Shanghai Municipal Commission of Commerce and director of the Expert Committee of Shanghai Online Shopping Chamber of Commerce,上海市商業(yè)經(jīng)濟學會會長齊曉齋,He Yunming, director and vice president of Lawson Supermarket, Yuan Bin, general manager of Feiniu.com, Feng Xuan, vice president of operations of Yu Ruifen of Shanghai Laiyifen Co., Ltd., and other big names in the retail industry all came.
As a leader in the retail passenger flow analysis industry in the business district, our company also made a grand appearance at this forum, and in this forum, our general manager, Mr. Xu Biao, was invited to conduct "Artificial Intelligence in Smart Business Circles". Application" keynote speech was affirmed by the professional audience and industry insiders.
As a leader in the retail passenger flow analysis industry in the business district, our company also made a grand appearance at this forum, and in this forum, our general manager, Mr. Xu Biao, was invited to conduct "Artificial Intelligence in Smart Business Circles". Application" keynote speech was affirmed by the professional audience and industry insiders.:Identification of customer attributes; refined data of passenger flow; thermal analysis of counters; night store deployment, and proposed comprehensive solutions for smart business districts.
The following is a summary of the speech delivered by Mr. Xu Biao, General Manager of Truthvision Technology:
非常榮幸有這么一個機會跟大家坐在一起進行交流,趨視科技是做人工智能,生產(chǎn)人工智能相關(guān)數(shù)據(jù)的公司,今天同大家做一些交流和探討,看趨視是不是可以在未來的商業(yè)零售生態(tài)里為大家提供更多的數(shù)據(jù)和參考。
我分四部分,第一部分是講一下現(xiàn)在商圈初期的智能化應用,可能客流多一點;第二闡述一下新技術(shù)的涌現(xiàn),包括AI技術(shù)涌現(xiàn)對商圈產(chǎn)生更豐富多元化的數(shù)據(jù);第三介紹一下要產(chǎn)生這樣的優(yōu)秀數(shù)據(jù)其實是依托于深入的AI底層數(shù)據(jù)的支撐,最后簡單介紹一下公司。
傳統(tǒng)客流分析
趨視對客流很關(guān)心,可能前幾年大家也了解了一些客流的數(shù)據(jù),可能想引入客流,了解我的商圈客流量有多少,包括每一層每一樓,包括每個鋪位來了多少人,這個是原來大家關(guān)注的,但是目前看來,簡單的來了多少人的客流數(shù)據(jù)并不是很有價值的商業(yè)數(shù)據(jù),很單一,并不能充分的對商圈和對未來的建設(shè)提供多元化的數(shù)據(jù)參考。第二,很多客流分析的安裝就是一個攝像頭,它各方面的應用上也不豐富,還存在很多數(shù)據(jù)的不準確,來了一萬人數(shù)成五千人這些等等,也帶不來豐富的應用,單純早期的客流在商圈的應用我們認為也到了一個升級換代的時候。
顧客屬性識別
二部分我們看一下新的AI在商圈中有那些應用。第一,精準的客流數(shù)據(jù),對客流要進行細分,這里面的細分包括來的人是誰,是不是你關(guān)注的人,初步來說,定義一個人到商店來有幾種,一種是店員,把它歸到客流里意義不大,因為他不是你所關(guān)注的。第二部分是VIP或者會員,這部分是大家關(guān)心的。第三部分既不是店員也不是會員,可能就是一個普通顧客,但是他昨天來過今天還來過,或者上個月來過,今天來過,這個人是不是有價值,這個數(shù)據(jù)獲取之后對后面的精準營銷可能會有幫助。對顧客的深度識別或者屬性化的識別,我們認為是對整個商圈數(shù)據(jù)挖掘的重要步驟。
柜臺熱力分析
另外柜臺的熱力圖,熱力圖是一種熱鬧程度,經(jīng)常來的柜臺熱力值自然就高,除了熱力圖之外還有顧客站在這里,待的時間是不是足夠長,對這個商品是不是關(guān)注了10分鐘以上,那對這個柜臺或者這個商品的定義就不一樣?;蛘邅淼娜撕芏啵沁M來之后30秒就走了,我們認為對客戶的行為分析也很重要,包括熱力圖數(shù)據(jù)看的是擁擠度和時長,這兩個數(shù)據(jù)很重要。攝像頭數(shù)據(jù)確定顧客的滯留時間,同時我們也可以用視頻數(shù)據(jù)做一個防盜。
動態(tài)人臉識別與行為分析算法
人臉的捕獲,一個人一個顏色代表是唯一一個人,這個人在走過程中可能有一張最好的人臉,我只關(guān)注他最好的人臉來識別他是誰,第一步的技術(shù)是準確的捕獲每一張人臉,這是人員的跟蹤,對一個人的整體行為,他在畫面中的坐標在哪里,軌跡怎么走的,每個人都會被完整的解釋出來,會被持續(xù)的跟蹤。也可以看到怎樣跟蹤到一個人,不管你攝像頭怎么架設(shè),角度對,都可以識別和跟蹤每個人的狀態(tài)。
這是綜合利用了現(xiàn)有在商圈和零售業(yè)里面的視頻資源,充分利用這些攝像機能夠產(chǎn)生多元化數(shù)據(jù),人看攝像機已經(jīng)沒有多大意義,而讓機器幫你去看攝像機,了解畫面中你想要的東西,白天的數(shù)據(jù)可能是客流的分析數(shù)據(jù)包括軌跡數(shù)據(jù),到夜間可能是監(jiān)控數(shù)據(jù)或者和些報警數(shù)據(jù),這些數(shù)據(jù)歸到上一層數(shù)據(jù)平臺,產(chǎn)生更多的數(shù)據(jù)交集。
最后講一下趨視科技,20年專注于人工智能核心技術(shù)的研發(fā)。致力于通過視覺行為分析技術(shù)與人臉識別應用,洞察消費者行為,為決策者提供精準而全面的商業(yè)信息,助力決策。