Recognizing human facial expressions is one of the most important and inspiring tasks in social communication.Face expressions are often a simple and obvious way for people to express their feelings and intentions.Facial expressions are the essential components of nonverbal communication.This study developed a model for detecting facial expressions that used both deep learning and andcrafted features. Preprocessing, feature extraction, and recognition are the three divisions of the suggested technique. The face recognition dataset is first preprocessed to reduce the noise using the Unsharp Masking Laplacian Non-linear Filter model.The shape, texture, and deep features are then retrieved from the preprocessed image using the Histogram of Oriented Gradients (HOG), Local Binary Pattern (LBP), and VGG16 network, respectively.These three features are also integrated to provide a more precise recognition. Finally, the K-Nearest Neighbour classifier is used to recognise the face expressions. The FER, CK+, and KDEF datasets are used to evaluate experiments. The suggested framework is successful and accurate, according to experimental results of seven emotion states (neutral, joy, sorrow, surprise, anger, fear, and disgust).

Deep Learning, Feature extraction, Facial expression recognition, Laplacian Non-Linear Filter, K Nearest Neighbour.