基于人脸属性识别广告精准投放智能新媒介系统设计

本论文结合高分子液晶透变薄膜、人脸识别和云计算技术,以商场透明隔离设施升级为一种智能新型媒介。首先,树莓派系统若检测到有人经过透明隔离设施后控制高分子透变膜使透明隔离设施分变成背投屏幕。然后系统采集人脸图像上传至云端并计算被测对象年龄、性别、表情等7项人脸属性信息,建立人脸数据库后云端返回人脸属性信息至系统。系统根据人脸属性结合消费场景推送广告,并检测游客的注视区域、时长等信息。最后将上述信息发送至云端与获取的人脸属性信息匹配生成闭环数据。经过一定的数据累积后,进行机器学习随机森林模型训练,挖掘客户消费能力、偏好趋向,进而实现广告精准投放。

This work combines the polymer liquid crystal translucent film,face recognition and cloud computing technology to turn the transparent isolation facility of the shopping mall into a new intelligent media.First of all,if the Raspberry Pi system detects that someone has passed through the transparent isolation facility,it controls the polymer permeable membrane to turn the transparent isolation facility into a backprojection screen.Then the system collects face images and uploads them to the cloud and calculates 7 facial attribute information such as the age,gender and expression of the tested object to establish a face database.After that,the cloud returns the facial attribute information to the system.The system pushes advertisements based on facial attributes and consumption scenes,and detects the information of tourists'gaze area and duration.Finally,the above information is sent to the cloud and matched with the acquired face attribute information to generate closed-loop data.After certain data accumulation,machine learning random forest model training is carried out to mine customers'consumption ability and preference trend,so as to achieve accurate advertising.