Design And Implementation Of Cloud-terminal-oriented Video Emotion Recognition System

The development of artificial intelligence technology has put forward higher requirements for human-computer interaction methods,and it is also possible for smart devices to better understand the user's emotional state.By analyzing the emotions of users,they can provide users with more personalized service strategies,which is of vital importance to improving service quality and user satisfaction.In power marketing,smart customer service,and other scenarios,the effect of emotion recognition is even more significant.The cloud-terminal-oriented video emotion recognition system is proposed in this context.This paper focuses on audio and video emotion recognition and combines fog computing environment to design and implement a system platform for audio and video emotion recognition.This paper proposes a video emotion recognition scheme based on deep neural networks,which uses facial data and facial key point data in combination with convolutional neural networks to complete the emotional recognition of video data.A two-way speech emotion recognition scheme based on CRNN is proposed.Through processing of the speech signal,extracting the statistical characteristics of the speech signal and features such as the Mel spectrum,it is trained using a recurrent neural network to achieve the emotion classification of the speech signal.For the audio-video emotion fusion scheme,in order to keep separate audio and video emotion recognition application scenario,a weighting criterion based audio and video emotion recognition fusion scheme is designed.Finally,in order to use the fog computing environment and reduce the delay caused by data transmission in the network,a cloud-terminal deployment scheme based on multi-server collaboration is proposed.Combined with the deployment scheme,the fog node completes the emotion recognition task processing,and the idea of coordinated control by the cloud computing center reduces data transmission in the network and achieves lower latency.Through the cloud-terminal-oriented video emotion recognition system,users can meet the needs of individual video emotion recognition,separate audio emotion recognition,and audio and video emotion recognition fusion.While ensuring high recognition accuracy,use fog nodes to complete emotion recognition tasks of edge devices and data are stored in the cloud computing center in a unified manner to achieve lower latency,reduce data transmission in the network,and reduce network bandwidth pressure.This article first introduces the research background for emotion recognition and the current status of research at home and abroad,and briefly introduces key technologies.Subsequently,on the basis of completing the requirements analysis,the key problems and solutions involved in the system are described in detail.Finally,the overall design and detailed design of the system were completed,including the overall architecture of the system,the design and implementation of the key modules of the system,etc.,and the system was tested as a whole to ensure that the functions met the requirements,and the availability of the system was verified.