The Design And Application Of Faical Data Collector Server In Mood-Recognition Video Monitoring Systems

With the development of society, the rhythm of work and life become faster, the pressure is also growing bigger. In this case, the number of psychological problems become more and more. In this reality, especially in some special places, such as schools, nursing homes and other places need to understand the psychological state of students, elderly by recognizing their emotion so that can provide psychological guidance for them. Therefore, we need a technology which can recognize the emotion by video to meet this demand. Mood-recognition video monitoring systems is a new application which can collect images from video and then use machine learning algorithms to analyze it to get photographer’s emotions.Mood-recognition video monitoring systems including facial data collect server, stream media server, the algorithm server, Web server and database server. Facial data collect server collects pictures from live video stream then send to algorithm server. Machine learning algorithms recognize facial emotion will take some time, so we choose a architecture which single-thread collect, multi-thread send and collect-thread share memory queue with send-thread. So that although machine learning algorithms recognize facial emotion will take some time, still be able to ensure the system’s throughput is maintained at a high level. Users both can watch facial emotion result which recognized by machine learnig algorithms and real-time video stream in terminal interface. Since the webcam only support limited output, when the number of users who view the video stream increase, in order to make sure that quality of collected images will not decline(the quality of collected images declining will affect emotion recognition), users should watch the live video forwarded by stream media server. Facial data collect server and stream media server both can accept request which sent by Web server and make a deal. As a server which run in the real environment, they both can recover from abnormal situation and provide services again.As described above, a facial data collect server which can maintain a higher throughput although machine learning algorithms recognize facial emotion will take some time and a stream media server which can forward Webcam’s live video are achieved in this paper. Test results show that facial data collect server and stream media server both reach the expected level.