Hybrid Features and Deep Learning Model for Facial Expression Recognition From Videos

Abstract

Facial expression recognition plays a crucial function in the advancement of technologies that can be used in detecting mental illness, sensors, and a wide variety of applications. Facial expression recognition is an interesting as well as strenuous task in digital field due to the complexity of the varying individuals. The intention of this work is to establish a face recognition model relying upon the modified GWO-based ensemble deep convolutional neural network (DCNN), which effectively recognizes the expressions. The substance of the research anticipates on the proposed modified GWO optimization which helps in maintaining the storage capacity with simple structures and provides high convergence. Enabling the optimization in the ensemble DCNN helps in tuning the internal parameters present in the classifier as well as helps in attaining best solution. The accomplishment of the proposed expression recognition model is evaluated utilizing the parameter metrics accuracy, precision, and recall that attained the values of 94.114%, 92.003%, and 95.734% which is more efficient.