Abstract
Facial video big sensor data (BSD) is the core data of wireless sensor network industry application and technology research. It plays an important role in many industries, such as urban safety management, unmanned driving, senseless attendance, and venue management. The construction of video big sensor data security application and intelligent algorithm model has become a hot and difficult topic in related fields based on facial expression recognition. This paper focused on the experimental analysis of Cohn–Kanade dataset plus (CK+) dataset with frontal pose and great clarity. Firstly, face alignment and the selection of peak image were utilized to preprocess the expression sequence. Then, the output vector from convolution network 1 and β-VAE were connected proportionally and input to support vector machine (SVM) classifier to complete facial expression recognition. The testing accuracy of the proposed model in CK + dataset can reach 99.615%. The number of expression sequences involved in training was 2417, and the number of expression sequences in testing was 519.
Highlights
Video is the best data type with largest amount and high degree of industrialization in big sensor data (BSD) applications
Facial expression recognition and analysis is the basic supporting technology of the above applications. It has become a hot and difficult topic to construct a security application and intelligent algorithm model of video BSD based on facial expression recognition [1, 2]. e main task of facial expression recognition is to realize automatic, reliable, and efficient facial information extraction and recognition
Turabzadeh et al achieved accuracy of 47.44% in automatic emotional detection by Field Programmable Gate Array (FPGA) [3]. en, Mehta et al realized emotion recognition in augmented reality (AR) by Microsoft HoloLens (MHL) [4]. e programming direction was aimed at optimizing the structure of model. ere were lots of datasets in the field of facial expression recognition, such as dataset Multi-PIE and CK+ [5, 6]
Summary
Video is the best data type with largest amount and high degree of industrialization in BSD applications. Zhang et al proposed an end-to-end deep learning model that utilized GAN network to achieve 91.8% accuracy on Multi-PIE dataset [9]. Face alignment and the selection of peak image were utilized to enhance data features. En, the output vector from convolution network 1 and β-VAE were connected proportionally and input to SVM classifier to complete facial expression recognition. E dataset has the disadvantages of small amount and irrelevant noise it has the above advantages and is preprocessed to improve the accuracy of facial expression recognition and the stability of training process. In each output of network, the coordinate of bbox corresponding to the highest value of score is intersected and calculated to find the IoU. Erefore, the structural similarity (SSIM) is utilized to calculate similarity to complete the selection of peak images [21]. After basic parameters are obtained, the three image comparison indices mentioned above are calculated, respectively, as shown in the following equations: Cov 3×3 Cov 3×3 Cov 3×3 MP 2×2
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
