智能數字標牌 : 觀眾表情分析
ART/166CP
20131231 - 20150629
10575
蔡振榮博士
A. Human expression recognition and analysis algorithm development (1) Static human Face expression recognition and analysis (a) 6 fundamental human expressions recognition for still image: Happiness, Sadness, Fear, Anger, Disgust and Surprise (b) > 80% accuracy (2) Dynamic human Face expression/emotion Analysis (a) Human emotion analysis for video sequence (b) > 80% accuracy B. Smart Digital Signage (1) Thin high-brightness LED backlight for digital signage (a) LED light-bar with 2nd optical lens for high-brightness backlight (>1500nit) with 40% light mixing distance reduction (b) Brightness: ≥1500 nits Tiled (2*2) LCD Size: ≥100”. (c) High resolution ambient adaptive power management (Power saving >50%) (2) 3D gesture recognition algorithm development for digital signage (a) Detection distance: 1 - 3 m (b) Detection functions: track and recognize a single user: gestures including clicking, dragging, sliding, rotating, enlarging and shrinking 3D objects (3) User ID verification by face recognition for digital signage (a) < 2% false accept rate; <3% false reject rate (b) Fixed acquisition/testing condition (angle, illumination, expression and decoration) C. Integrated Audience Analysis System (1) Low-cost Intelligent Audience Analysis (IAA) (a) People counting and watch time recording (b) > 90% accuracy (2) High performance version (a) > 90% accuracy in people counting and watch time recording (b) > 80% accuracy in expression, age and gender classification (c) Near-frontal-view face (left-right:30 degree). (d) Detection Distance: Face Detection and Tracking: 0.3~6m, Age , Gender and expression:1~3m.
Dr Enboa WU
Ms Ying, Anna LIU
Mr Kin Lung, Kenny CHAN
Dr Xiu Ling ZHU
Dr Zhao, Fiona Wang
Mr Chang-Li WU
Dr Chi Yong
Mr Yuk lung, Alex Cheung
Mr Wei-Ping Tang
Ms Wei, Wendy Zhang
Dr Yan Wang
Mr Derek Dehua Liu
Dr Xueyan TANG
Mr Chun Yip Wong
Mr Chan Kwok Chung
Mr Dennis Chi Kin Wong
Mr Wai Kin Luk
Mr Pei Chin Ryan Chung
Mr Weiwen, Wilman Zou
Mr Hui Hu
Mr Haocheng Han
Dr Mi Suen Lee
Ms Jie, Fiona Lu
Dr Jie Kou
Mr Guanghua Huang
Chinese Academy of Science (CAS)
香港榮立科技有限公司
HK Ronglee Technology Limited (Technology Licensing) [贊助機構]
深圳穗晶光電有限公司 [贊助機構]
The University of Nottingham
震威電子有限公司
蕪湖銳拓電子有限公司 [贊助機構]
根據市場調查分析報告顯示全球的電子廣告業(digital advertising) 已經步入了飛速發展的階段。現時的廣告行業仍然採用傳統方式，單向傳遞資訊給用戶，而忽視了用戶的訊息、反饋與互動。如果能夠實現傳遞給用戶感興趣的資訊，智慧數字標牌將會跨入一個嶄新的發展階段，然而目前的市場中並沒有成熟的智慧數字標牌的產品。 因此本項目主要目的在於開發的新一代智慧數字標牌技術與產品。為開發此智能數字標牌技術與產品，本項目將專注下列技術平臺的開發，包括(1). 觀眾表情識別: 通過檢測用戶的面部表情,進行觀眾喜好分析及或進一步提供用戶感興趣的資訊; (2). 非接觸式手勢互動控制: 讓用戶通過觸摸和遠: 程手勢兩種方式來控制顯示資訊; (3). 整合嵌入觀眾分析系統: 整合人臉識別、關注程度、年齡、性別及表情喜好分析系統及根據產品需求開發適合嵌入式模組。