基於智慧眼鏡偵測使用者偏好之互動回饋學習
Owing to the wide availability of wearable devices such as smart glasses,the impact brings the proliferation of applications in various domains.In this paper,we utilize sensor-rich smart glasses to collect multiple information about a user including the user's position,viewing direction,and attention information when viewing exhibition objects in a certain location.The intent is to analyze the user's preferences from the user with smart glasses as the interactive feedbacks for a learning system.Therefore,we propose the user preference detection(UPD)method to find the user's interesting point.For this purpose,we first propose a 3D key fragment extraction method to obtain important frames for the user.We filter out unimportant frames which include the user walking,turning around quickly,and high or low viewing angle scenes.In addition,we utilize the attention level with PPG sensors to accurately select key frames.Second,we detect a point of interest(POI)that a user pays attention to by performing a personalized POI detection method which utilizes a personal profile to acquire the most possible POI for the user.We utilize the distance-based normal distribution model and user's gazing point correction to obtain potential POIs.Subsequently,we adopt the personal profile to calculate the POI score and find the POI with the highest score to represent the user interesting point for the recommendation system.Finally,we present the experimental results on data sets with various distributions to demonstrate the performance and utility of our approach.
wearable devises e-learning smart learning POI detection smart glasses recommendation systems
Wen-Ren Lin Yu-Ling Hsueh Jerry Chih-Yuan Sun
国内会议
武汉
中文
247-252
2019-05-23（万方平台首次上网日期，不代表论文的发表时间）