Facial Expression Recognition Research Articles

Automatic facial expression localization and recognition across a large range of emotions

Automatic facial expression localization and recognition across a large range of emotions

Recognition and classification of facial expression using artificial intelligence as a key of early detection in neurological disorders.

The recognition and classification of facial expressions using artificial intelligence (AI) presents a promising avenue for early detection and monitoring ofneurodegenerative disorders. This narrative review critically examines the current state of AI-driven facial expression analysis in the context of neurodegenerative diseases, such as Alzheimer's and Parkinson's. We discuss thepotential of AI techniques, including deep learning andcomputer vision, to accurately interpret and categorize subtle changes in facial expressions associated with thesepathological conditions. Furthermore, we explore theroleof facial expression recognition as a noninvasive, cost-effective tool for screening, disease progression tracking, and personalized intervention in neurodegenerative disorders. The review also addresses the challenges, ethical considerations, and future prospects of integrating AI-based facial expression analysis into clinical practice forearly intervention and improved quality of life for individuals at risk ofor affected by neurodegenerative diseases.

Real-Time Facial Expression Recognition Based on Image Processing in Virtual Reality

More virtual reality (VR) scenarios have become more prevalent in recent years. More and more people are getting into VR, meaning that objective physiological measures to assess a user's emotional state automatically are becoming more critical. Individuals’ emotional states impact their behaviour, opinions, emotions, and decisions. They may be used to analyze VR experiences and make systems react to and engage with the user’s emotions. VR environments require users to wear head-mounted displays (HMDs), blocking off their upper faces. That makes traditional Facial Expression Recognition (FER) approaches very limited in their usefulness. Thus, a Deep Learning (DL) solution combined with image processing is utilized to classify universal emotions: sadness, happiness, disgust, anger, fear and surprise. Hence, this paper suggests the Deep Automatic Facial Expression Recognition Model (DAFERM) for interactive virtual reality (VR) applications such as intelligent education, social networks, and virtual training. Two main parts comprise the system: one that uses deep neural networks (DNNs) for facial emotion identification and another that automatically tracks and segments faces. The system begins by following a marker on the front of the head-mounted display (HMD). With the help of the spatial data that has been retrieved, the positions and rotations of the face are estimated to segment the mouth. Finally, the system interacts with DNN using the pixels processed by the lips. It obtains the facial expression results in real time using an adaptive method for histogram-based mouth segmentation.

Categorization processes in emotion expression recognition: The roles of language and essentialism.

We examined categorical processing biases in the perception and recognition of facial expressions of emotion across two studies. In both studies, participants first learned to discriminate between two ambiguous facial expressions of emotion selected from the middle of a continuous array of blended expressions (i.e., an array created from morphing images of two facial expressions together, with still images selected in equidistant increments). Participants were then asked to recognize the specific expressions they were trained to discriminate. In Study 1, target expressions labeled with emotion words (e.g., more angry face) during discrimination and recognition tasks were misremembered as more perceptually distinct from one another and therefore more perceptually similar to the stereotypical expression for their labeled emotion category than they were in reality. Critically, in Study 2, these recognition biases were significantly reduced or absent in conditions where the target expressions were not labeled with emotion words (e.g., Face A), demonstrating the role of emotion words in promoting categorical processing biases in emotion recognition. Moreover, in the absence of emotion labels, peoples' beliefs about the nature of emotion categories were related to the extent to which they employed categorical processing during emotion perception and recognition. Specifically, people with more essentialist beliefs about emotion categories-believing emotion categories are more innate, biologically-based, and immutable-exhibited more pronounced categorical processing biases during emotion recognition. Findings shed light on the critical role of language and cognition in constructing emotion and add to empirical findings on categorical processing in emotion perception. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

A facial expression recognition network using hybrid feature extraction.

Facial expression recognition faces great challenges due to factors such as face similarity, image quality, and age variation. Although various existing end-to-end Convolutional Neural Network (CNN) architectures have achieved good classification results in facial expression recognition tasks, these network architectures share a common drawback that the convolutional kernel can only compute the correlation between elements of a localized region when extracting expression features from an image. This leads to difficulties for the network to explore the relationship between all the elements that make up a complete expression. In response to this issue, this article proposes a facial expression recognition network called HFE-Net. In order to capture the subtle changes of expression features and the whole facial expression information at the same time, HFE-Net proposed a Hybrid Feature Extraction Block. Specifically, Hybrid Feature Extraction Block consists of parallel Feature Fusion Device and Multi-head Self-attention. Among them, Feature Fusion Device not only extracts the local information in expression features, but also measures the correlation between distant elements in expression features, which helps the network to focus more on the target region while realizing the information interaction between distant features. And Multi-head Self-attention can calculate the correlation between the overall elements in the feature map, which helps the network to extract the overall information of the expression features. We conducted a lot of experiments on four publicly available facial expression datasets and verified that the Hybrid Feature Extraction Block constructed in this paper can improve the network's recognition ability for facial expressions.

Enhancing Facial Expression Recognition with Robust CNN Architectures and Adaptive Preprocessing Techniques

Facial expression recognition (FER) is an essential technology at the intersection of artificial intelligence (AI), computer vision, and psychology. This study proposes a novel framework for FER, aiming to improve system robustness and generalization, especially under variable real-world conditions. Using the FER2013 dataset, this research combines an adaptive preprocessing pipeline with a custom Convolutional Neural Network (CNN) architecture. Key preprocessing steps include normalization, rotation, and flipping to improve data quality and diversity. The CNN architecture combines regularization methods, including dropout and L2 regularization. Dynamic hyperparameter tuning and early stopping optimize performance and prevent overfitting. Normalized confusion matrix indicating strong recognition for well-represented emotions, such as happiness with 86% accuracy, and challenges with underrepresented categories like disgust. This research aims to contribute to the ongoing development of facial expression recognition systems by enhancing their robustness and generalization. While further refinement is needed, this work provides a step toward more accurate and adaptable FER models, with the potential to support advancements in human-computer interaction and various real-world applications.

Ultrasensitive iontronic pressure sensor based on microstructure ionogel dielectric layer for wearable electronics

Flexible pressure sensors show great promise for applications in such fields as electronic skin, healthcare, and intelligent robotics. Traditional capacitive pressure sensors, however, face the problem of low sensitivity, which limits their wider application. In this paper, a flexible capacitive pressure sensor with microstructured ionization layer is fabricated by a sandwich-type process, with a low-cost and simple process of inverted molding with sandpapers being used to form a thermoplastic polyurethane elastomer ionic film with double-sided microstructure as the dielectric layer of the sensor, with silver nanowires as electrodes. The operating mechanism of this iontronic pressure sensor is analyzed using a graphical method, and the sensor is tested on a pressure platform. The test results show that the sensor has ultrahigh pressure sensitivities of 3.744 and 1.689 kPa−1 at low (0–20 kPa) and high (20–800 kPa) pressures, respectively, as well as a rapid response time (100 ms), and it exhibits good stability and repeatability. The sensor can be used for sensitive monitoring of activities such as finger bending, and for facial expression (smile, frown) recognition, as well as speech recognition.

A joint learning method for low-light facial expression recognition

Existing facial expression recognition (FER) methods are mainly devoted to learning discriminative features from normal-light images. However, their performance drops sharply when they are used for low-light images. In this paper, we propose a novel low-light FER framework (termed LL-FER) that can simultaneously enhance the images and recognition tasks of low-light facial expression images. Specifically, we first meticulously design a low-light enhancement network (LLENet) to recover expressions images’ rich detail information. Then, we design a joint loss to train the LLENet with FER network in a cascade manner, so that the FER network can guide the LLENet to gradually perceive and restore discriminative features which are useful for FER during the training process. Extensive experiments show that the LLENet not only achieves competitive results both quantitatively and qualitatively, but also in the LL-FER framework, which can produce results more suitable for FER tasks, further improving the performance of the FER methods.

Effects of repeated intravenous esketamine administration on affective biases

Objectives While neuropsychological effects of conventional antidepressants are well-documented, more research is needed for rapid-acting antidepressants. This study examines the effects of esketamine on emotion processing and cognitive functioning, both acutely and sub-chronically. Methods Eighteen treatment-resistant depression (TRD) patients received repeated intravenous esketamine infusions. Mood state was reported daily, and the Facial Expression Recognition Task was administered 1h before and 4h after each infusion. Other assessments included the Digit Symbol Substitution Task. Results 66.7% participants who received at least five infusions (n = 12) showed significant improvement. Emotion recognition improved for all emotions except sadness, where accuracy decreased, particularly for low-intensity expressions (p = .007, d = −1.09). Misclassifications of other emotions as sad also decreased (p = .035, d = −0.79), indicating a reduced response bias towards sadness. This shift in bias emerged after the first infusion and then consolidated over time. In parallel, participants showed significant reductions in feelings of sadness (p = .015, d = −0.89) and irritability (p = .001, d = −1.35). Symptomatic improvement negatively correlated with accuracy for and misclassifications of sadness, and cognitive functioning also improved (p = .001, d = 1.62). Conclusions Improvement of TRD by esketamine may involve shifts in emotion processing and cognition, with the acute mood-lifting effects of esketamine being discernible from longer-lasting antidepressant response, which consolidates after repeated administration.

Facial emotion recognition based on upper features and transfer learning

Facial expression recognition (FER) in the upper face focuses on the analysis and recognition of emotions based on features extracted from the upper region of the face. This region typically affects the eyes, eyebrows, forehead, and sometimes the upper cheeks. Since these areas are often less affected by face masks or other facial coverings, FER algorithms can concentrate on capturing and interpreting the relevant facial cues, such as eye movements, eyebrow positions, and forehead wrinkles, to accurately recognize and classify different emotions. By focusing on the upper face, FER systems can mitigate the impact of occlusions caused by masks and still provide meaningful insights into the emotional states of individuals. In this work, a FER approach focusing on the upper region is proposed. Several experiments have been made using the CK+ dataset in addition to a comparison between the emotion recognition scores using the upper and the entire face in order to determine whether this area can reflect the real expressed emotion. The results of our approach are promising compared to previous studies with an accuracy up to 96%.

Mixture of Emotion Dependent Experts: Facial Expressions Recognition in Videos through Stacked Expert Models

Mixture of Emotion Dependent Experts: Facial Expressions Recognition in Videos through Stacked Expert Models

Heterogeneous Dual-Branch Emotional Consistency Network for Facial Expression Recognition

Heterogeneous Dual-Branch Emotional Consistency Network for Facial Expression Recognition

Facial expression recognition based on YOLOv8 deep learning in complex scenes

Facial expression recognition based on YOLOv8 deep learning in complex scenes

Self‐learning weight network based on label distribution training for facial expression recognition

Self‐learning weight network based on label distribution training for facial expression recognition

Global multi-scale extraction and local mixed multi-head attention for facial expression recognition in the wild

Global multi-scale extraction and local mixed multi-head attention for facial expression recognition in the wild

Co-dance with Ambiguity: An Ambiguity-Aware Facial Expression Recognition Framework for More Robustness

Co-dance with Ambiguity: An Ambiguity-Aware Facial Expression Recognition Framework for More Robustness

Unsupervised Cross-Domain Facial Expression Recognition via Class Adaptive self-Training

Unsupervised Cross-Domain Facial Expression Recognition via Class Adaptive self-Training

Enhanced Autism Spectrum Disorder Facial Expression Recognition Using Hybrid Weighed Quantum Particle Swarm Optimization with Fast Mask Recurrent Convolutional Neural Network

Enhanced Autism Spectrum Disorder Facial Expression Recognition Using Hybrid Weighed Quantum Particle Swarm Optimization with Fast Mask Recurrent Convolutional Neural Network

Time Facial Expression Recognition Using Optimized CNN Models for Behavioral and Emotional Analysis

Time Facial Expression Recognition Using Optimized CNN Models for Behavioral and Emotional Analysis

Facial expression recognition with modified local ternary pattern images using convolutional neural network and extreme learning machine

Facial emotion recognition is a crucial aspect of human-computer interaction systems, education, health and various other fields. The common challenges faced in FER are variation in light. Based on researches, Local ternary Pattern (LTP) as a feature vector that can handle the variation of light. Modified Local Ternary Pattern (MLTP) is more discriminative to variation of light and less sensitive to noise. MLTP is a conventional feature vector that requires manual processing. In contrast to MLTP, the Convolution Neural Network (CNN) architecture has an automated feature extractor. This paper proposes MLTP as input to the CNN architecture for handling the variation of light and for automatic feature extractor. Then, for classifying Extreme Learning Machine is used to reduce the training time of the CNN classifier. The Karolinska Directed Emotional Faces (KDEF) dataset is used to assess the proposed method and got an accuracy of 86.28%.