Facial expressions are a powerful medium for conveying emotions. In facial expression recognition (FER) field, the difficulty of collecting specific expressions often leads to class imbalance in mainstream datasets, significantly reducing the classification accuracy of deep neural networks. To address these issues, we propose a stable-diffusion-based augmentation method for facial expression (SDAFE) that resolves class imbalance problems and enhances data generation quality through cross-modal label guidance. By leveraging the neutrality of neutral faces, we generate additional expressions to balance the dataset classes. We introduce a peak signal-to-noise ratio (PSNR) filter to ensure the high quality of the generated images and a cosine similarity cross-modal filter based on CLIP encoders to ensure that the content of the generated images accurately aligns with their labels. Furthermore, we introduce a novel model, FERNeXt, which demonstrates outstanding performance in FER tasks, surpassing the state-of-the-art accuracy on the FER2013 dataset and achieving strong results on the RAF-DB and NHFI datasets. Subsequently, the performance of several models across different datasets significantly improves through the use of SDAFE in our experiments.
Prompt learning has been widely adopted to efficiently adapt vision-language models (VLMs) like CLIP for various downstream tasks. Despite their success, current VLM-based facial expression recognition (FER) methods struggle to capture fine-grained textual-visual relationships, which are essential for distinguishing subtle differences between facial expressions. To address this challenge, we propose a multimodal prompt alignment framework for FER, called MPA-FER, that provides fine-grained semantic guidance to the learning process of prompted visual features, resulting in more precise and interpretable representations. Specifically, we introduce a multi-granularity hard prompt generation strategy that utilizes a large language model (LLM) like ChatGPT to generate detailed descriptions for each facial expression. The LLM-based external knowledge is injected into the soft prompts by minimizing the feature discrepancy between the soft prompts and the hard prompts. To preserve the generalization abilities of the pretrained CLIP model, our approach incorporates prototype-guided visual feature alignment, ensuring that the prompted visual features from the frozen image encoder align closely with class-specific prototypes. Additionally, we propose a cross-modal global-local alignment module that focuses on expression-relevant facial features, further improving the alignment between textual and visual features. Extensive experiments demonstrate our framework outperforms state-of-the-art methods on three FER benchmark datasets, while retaining the benefits of the pretrained model and minimizing computational costs.
Dynamic facial expression recognition (DFER) is a task that estimates emotions from facial expression video sequences. For practical applications, accurately recognizing ambiguous facial expressions -- frequently encountered in in-the-wild data -- is essential. In this study, we propose MIDAS, a data augmentation method designed to enhance DFER performance for ambiguous facial expression data using soft labels representing probabilities of multiple emotion classes. MIDAS augments training data by convexly combining pairs of video frames and their corresponding emotion class labels. This approach extends mixup to soft-labeled video data, offering a simple yet highly effective method for handling ambiguity in DFER. To evaluate MIDAS, we conducted experiments on both the DFEW dataset and FERV39k-Plus, a newly constructed dataset that assigns soft labels to an existing DFER dataset. The results demonstrate that models trained with MIDAS-augmented data achieve superior performance compared to the state-of-the-art method trained on the original dataset.
Students' academic emotions significantly influence their social behavior and learning performance. Traditional approaches to automatically and accurately analyze these emotions have predominantly relied on supervised machine learning algorithms. However, these models often struggle to generalize across different contexts, necessitating repeated cycles of data collection, annotation, and training. The emergence of Vision-Language Models (VLMs) offers a promising alternative, enabling generalization across visual recognition tasks through zero-shot prompting without requiring fine-tuning. This study investigates the potential of VLMs to analyze students' academic emotions via facial expressions in an online learning environment. We employed two VLMs, Llama-3.2-11B-Vision-Instruct and Qwen2.5-VL-7B-Instruct, to analyze 5,000 images depicting confused, distracted, happy, neutral, and tired expressions using zero-shot prompting. Preliminary results indicate that both models demonstrate moderate performance in academic facial expression recognition, with Qwen2.5-VL-7B-Instruct outperforming Llama-3.2-11B-Vision-Instruct. Notably, both models excel in identifying students' happy emotions but fail to detect distracted behavior. Additionally, Qwen2.5-VL-7B-Instruct exhibits relatively high performance in recognizing students' confused expressions, highlighting its potential for practical applications in identifying content that causes student confusion.
Facial expression recognition (FER), aiming to classify human emotions automatically, is a significant problem in computer vision. Recent advancements in deep learning and computer vision have led to notable progress in FER. This work proposes an enhanced emotion recognition framework utilizing the FER-2013 dataset, augmented with additional training data for improved generalization performance. The EfficientNetv2 architecture is employed with transfer learning for robust and comprehensive feature extraction. The proposed method leverages attention mechanisms to capture critical facial details while mitigating the influence of irrelevant information. The model trained with approximately 23.8 million parameters surpassed the performance of existing methods by classifying with %82.47 accuracy rate on the FER-2013 dataset. These results indicate the potential applicability of the proposed approach to emotion recognition tasks.
Facial expression recognition (FER) in the wild remains a challenging task due to the subtle and localized nature of expression-related features, as well as the complex variations in facial appearance. In this paper, we introduce a novel framework that explicitly focuses on Texture Key Driver Factors (TKDF), localized texture regions that exhibit strong discriminative power across emotional categories. By carefully observing facial image patterns, we identify that certain texture cues, such as micro-changes in skin around the brows, eyes, and mouth, serve as primary indicators of emotional dynamics. To effectively capture and leverage these cues, we propose a FER architecture comprising a Texture-Aware Feature Extractor (TAFE) and Dual Contextual Information Filtering (DCIF). TAFE employs a ResNet-based backbone enhanced with multi-branch attention to extract fine-grained texture representations, while DCIF refines these features by filtering context through adaptive pooling and attention mechanisms. Experimental results on RAF-DB and KDEF datasets demonstrate that our method achieves state-of-the-art performance, verifying the effectiveness and robustness of incorporating TKDFs into FER pipelines.
In this paper, we introduce MultiviewVLM, a vision-language model designed for unsupervised contrastive multiview representation learning of facial emotions from 3D/4D data. Our architecture integrates pseudo-labels derived from generated textual prompts to guide implicit alignment of emotional semantics. To capture shared information across multi-views, we propose a joint embedding space that aligns multiview representations without requiring explicit supervision. We further enhance the discriminability of our model through a novel multiview contrastive learning strategy that leverages stable positive-negative pair sampling. A gradient-friendly loss function is introduced to promote smoother and more stable convergence, and the model is optimized for distributed training to ensure scalability. Extensive experiments demonstrate that MultiviewVLM outperforms existing state-of-the-art methods and can be easily adapted to various real-world applications with minimal modifications.
Facial expression recognition (FER) remains a challenging task due to label ambiguity caused by the subjective nature of facial expressions and noisy samples. Additionally, class imbalance, which is common in real-world datasets, further complicates FER. Although many studies have shown impressive improvements, they typically address only one of these issues, leading to suboptimal results. To tackle both challenges simultaneously, we propose a novel framework called Navigating Label Ambiguity (NLA), which is robust under real-world conditions. The motivation behind NLA is that dynamically estimating and emphasizing ambiguous samples at each iteration helps mitigate noise and class imbalance by reducing the model's bias toward majority classes. To achieve this, NLA consists of two main components: Noise-aware Adaptive Weighting (NAW) and consistency regularization. Specifically, NAW adaptively assigns higher importance to ambiguous samples and lower importance to noisy ones, based on the correlation between the intermediate prediction scores for the ground truth and the nearest negative. Moreover, we incorporate a regularization term to ensure consistent latent distributions. Consequently, NLA enables the model to progressively focus on more challenging ambiguous samples, which primarily belong to the minority class, in the later stages of training. Extensive experiments demonstrate that NLA outperforms existing methods in both overall and mean accuracy, confirming its robustness against noise and class imbalance. To the best of our knowledge, this is the first framework to address both problems simultaneously.
The emergence of ConvNeXt and its variants has reaffirmed the conceptual and structural suitability of CNN-based models for vision tasks, re-establishing them as key players in image classification in general, and in facial expression recognition (FER) in particular. In this paper, we propose a new set of models that build on these advancements by incorporating a new set of attention mechanisms that combines Triplet attention with Squeeze-and-Excitation (TripSE) in four different variants. We demonstrate the effectiveness of these variants by applying them to the ResNet18, DenseNet and ConvNext architectures to validate their versatility and impact. Our study shows that incorporating a TripSE block in these CNN models boosts their performances, particularly for the ConvNeXt architecture, indicating its utility. We evaluate the proposed mechanisms and associated models across four datasets, namely CIFAR100, ImageNet, FER2013 and AffectNet datasets, where ConvNext with TripSE achieves state-of-the-art results with an accuracy of \textbf{78.27\%} on the popular FER2013 dataset, a new feat for this dataset.
Facial Expression Recognition (FER) plays a foundational role in enabling AI systems to interpret emotional nuances, a critical aspect of affective Theory of Mind (ToM). However, existing models often struggle with poor calibration and a limited capacity to capture emotional intensity and complexity. To address this, we propose Ranking the Emotional Nuance for Theory of Mind (Rank-O-ToM), a framework that leverages ordinal ranking to align confidence levels with the emotional spectrum. By incorporating synthetic samples reflecting diverse affective complexities, Rank-O-ToM enhances the nuanced understanding of emotions, advancing AI's ability to reason about affective states.
We study whether and how the choice of optimization algorithm can impact group fairness in deep neural networks. Through stochastic differential equation analysis of optimization dynamics in an analytically tractable setup, we demonstrate that the choice of optimization algorithm indeed influences fairness outcomes, particularly under severe imbalance. Furthermore, we show that when comparing two categories of optimizers, adaptive methods and stochastic methods, RMSProp (from the adaptive category) has a higher likelihood of converging to fairer minima than SGD (from the stochastic category). Building on this insight, we derive two new theoretical guarantees showing that, under appropriate conditions, RMSProp exhibits fairer parameter updates and improved fairness in a single optimization step compared to SGD. We then validate these findings through extensive experiments on three publicly available datasets, namely CelebA, FairFace, and MS-COCO, across different tasks as facial expression recognition, gender classification, and multi-label classification, using various backbones. Considering multiple fairness definitions including equalized odds, equal opportunity, and demographic parity, adaptive optimizers like RMSProp and Adam consistently outperform SGD in terms of group fairness, while maintaining comparable predictive accuracy. Our results highlight the role of adaptive updates as a crucial yet overlooked mechanism for promoting fair outcomes.
The human face plays a central role in social communication, necessitating the use of performant computer vision tools for human-centered applications. We propose Face-LLaVA, a multimodal large language model for face-centered, in-context learning, including facial expression and attribute recognition. Additionally, Face-LLaVA is able to generate natural language descriptions that can be used for reasoning. Leveraging existing visual databases, we first developed FaceInstruct-1M, a face-centered database for instruction tuning MLLMs for face processing. We then developed a novel face-specific visual encoder powered by Face-Region Guided Cross-Attention that integrates face geometry with local visual features. We evaluated the proposed method across nine different datasets and five different face processing tasks, including facial expression recognition, action unit detection, facial attribute detection, age estimation and deepfake detection. Face-LLaVA achieves superior results compared to existing open-source MLLMs and competitive performance compared to commercial solutions. Our model output also receives a higher reasoning rating by GPT under a zero-shot setting across all the tasks. Both our dataset and model wil be released at https://face-llava.github.io to support future advancements in social AI and foundational vision-language research.
Robust facial expression recognition in unconstrained, "in-the-wild" environments remains challenging due to significant domain shifts between training and testing distributions. Test-time adaptation (TTA) offers a promising solution by adapting pre-trained models during inference without requiring labeled test data. However, existing TTA approaches typically rely on manually selecting which parameters to update, potentially leading to suboptimal adaptation and high computational costs. This paper introduces a novel Fisher-driven selective adaptation framework that dynamically identifies and updates only the most critical model parameters based on their importance as quantified by Fisher information. By integrating this principled parameter selection approach with temporal consistency constraints, our method enables efficient and effective adaptation specifically tailored for video-based facial expression recognition. Experiments on the challenging AffWild2 benchmark demonstrate that our approach significantly outperforms existing TTA methods, achieving a 7.7% improvement in F1 score over the base model while adapting only 22,000 parameters-more than 20 times fewer than comparable methods. Our ablation studies further reveal that parameter importance can be effectively estimated from minimal data, with sampling just 1-3 frames sufficient for substantial performance gains. The proposed approach not only enhances recognition accuracy but also dramatically reduces computational overhead, making test-time adaptation more practical for real-world affective computing applications.
Facial Expression Recognition (FER) from videos is a crucial task in various application areas, such as human-computer interaction and health monitoring (e.g., pain, depression, fatigue, and stress). Beyond the challenges of recognizing subtle emotional or health states, the effectiveness of deep FER models is often hindered by the considerable variability of expressions among subjects. Source-free domain adaptation (SFDA) methods are employed to adapt a pre-trained source model using only unlabeled target domain data, thereby avoiding data privacy and storage issues. Typically, SFDA methods adapt to a target domain dataset corresponding to an entire population and assume it includes data from all recognition classes. However, collecting such comprehensive target data can be difficult or even impossible for FER in healthcare applications. In many real-world scenarios, it may be feasible to collect a short neutral control video (displaying only neutral expressions) for target subjects before deployment. These videos can be used to adapt a model to better handle the variability of expressions among subjects. This paper introduces the Disentangled Source-Free Domain Adaptation (DSFDA) method to address the SFDA challenge posed by missing target expression data. DSFDA leverages data from a neutral target control video for end-to-end generation and adaptation of target data with missing non-neutral data. Our method learns to disentangle features related to expressions and identity while generating the missing non-neutral target data, thereby enhancing model accuracy. Additionally, our self-supervision strategy improves model adaptation by reconstructing target images that maintain the same identity and source expression. Experimental results on the challenging BioVid and UNBC-McMaster pain datasets indicate that our DSFDA approach can outperform state-of-the-art adaptation method.
This study investigates the key characteristics and suitability of widely used Facial Expression Recognition (FER) datasets for training deep learning models. In the field of affective computing, FER is essential for interpreting human emotions, yet the performance of FER systems is highly contingent on the quality and diversity of the underlying datasets. To address this issue, we compiled and analyzed 24 FER datasets, including those targeting specific age groups such as children, adults, and the elderly, and processed them through a comprehensive normalization pipeline. In addition, we enriched the datasets with automatic annotations for age and gender, enabling a more nuanced evaluation of their demographic properties. To further assess dataset efficacy, we introduce three novel metricsLocal, Global, and Paired Similarity, which quantitatively measure dataset difficulty, generalization capability, and cross-dataset transferability. Benchmark experiments using state-of-the-art neural networks reveal that large-scale, automatically collected datasets (e.g., AffectNet, FER2013) tend to generalize better, despite issues with labeling noise and demographic biases, whereas controlled datasets offer higher annotation quality but limited variability. Our findings provide actionable recommendations for dataset selection and design, advancing the development of more robust, fair, and effective FER systems.
Vision-Language Models (VLMs) like CLIP offer promising solutions for Dynamic Facial Expression Recognition (DFER) but face challenges such as inefficient full fine-tuning, high complexity, and poor alignment between textual and visual representations. Additionally, existing methods struggle with ineffective temporal modeling. To address these issues, we propose PE-CLIP, a parameter-efficient fine-tuning (PEFT) framework that adapts CLIP for DFER while significantly reducing trainable parameters while maintaining high accuracy. PE-CLIP introduces two specialized adapters: a Temporal Dynamic Adapter (TDA) and a Shared Adapter (ShA). The TDA is a GRU-based module with dynamic scaling that captures sequential dependencies while emphasizing informative temporal features and suppressing irrelevant variations. The ShA is a lightweight adapter that refines representations within both textual and visual encoders, ensuring consistency and efficiency. Additionally, we integrate Multi-modal Prompt Learning (MaPLe), introducing learnable prompts for visual and action unit-based textual inputs, enhancing semantic alignment between modalities and enabling efficient CLIP adaptation for dynamic tasks. We evaluate PE-CLIP on two benchmark datasets, DFEW and FERV39K, achieving competitive performance compared to state-of-the-art methods while requiring fewer trainable parameters. By balancing efficiency and accuracy, PE-CLIP sets a new benchmark in resource-efficient DFER. The source code of the proposed PE-CLIP will be publicly available at https://github.com/Ibtissam-SAADI/PE-CLIP .
Semi-supervised deep facial expression recognition (SS-DFER) has gained increasingly research interest due to the difficulty in accessing sufficient labeled data in practical settings. However, existing SS-DFER methods mainly utilize generated semantic-level pseudo-labels for supervised learning, the unreliability of which compromises their performance and undermines the practical utility. In this paper, we propose a novel SS-DFER framework that simultaneously incorporates semantic, instance, and text-level information to generate high-quality pseudo-labels. Specifically, for the unlabeled data, considering the comprehensive knowledge within the textual descriptions and instance representations, we respectively calculate the similarities between the facial vision features and the corresponding textual and instance features to obtain the probabilities at the text- and instance-level. Combining with the semantic-level probability, these three-level probabilities are elaborately aggregated to gain the final pseudo-labels. Furthermore, to enhance the utilization of one-hot labels for the labeled data, we also incorporate text embeddings excavated from textual descriptions to co-supervise model training, enabling facial visual features to exhibit semantic correlations in the text space. Experiments on three datasets demonstrate that our method significantly outperforms current state-of-the-art SS-DFER methods and even exceeds fully supervised baselines. The code will be available at https://github.com/PatrickStarL/SIT-FER.
Facial Expression Recognition (FER) plays a crucial role in human affective analysis and has been widely applied in computer vision tasks such as human-computer interaction and psychological assessment. The 8th Affective Behavior Analysis in-the-Wild (ABAW) Challenge aims to assess human emotions using the video-based Aff-Wild2 dataset. This challenge includes various tasks, including the video-based EXPR recognition track, which is our primary focus. In this paper, we demonstrate that addressing label ambiguity and class imbalance, which are known to cause performance degradation, can lead to meaningful performance improvements. Specifically, we propose Video-based Noise-aware Adaptive Weighting (V-NAW), which adaptively assigns importance to each frame in a clip to address label ambiguity and effectively capture temporal variations in facial expressions. Furthermore, we introduce a simple and effective augmentation strategy to reduce redundancy between consecutive frames, which is a primary cause of overfitting. Through extensive experiments, we validate the effectiveness of our approach, demonstrating significant improvements in video-based FER performance.
Facial expression recognition is a challenging classification task with broad application prospects in the field of human - computer interaction. This paper aims to introduce the methods of our upcoming 8th Affective Behavior Analysis in the Wild (ABAW) competition to be held at CVPR2025. To address issues such as low recognition accuracy caused by subtle expression changes and multi - scales in facial expression recognition in videos, we propose global channel - spatial attention and median - enhanced spatial - channel attention to strengthen feature processing for speech and images respectively. Secondly, to fully utilize the complementarity between the speech and facial expression modalities, a speech - and - facial - expression key - frame alignment technique is adopted to calculate the weights of speech and facial expressions. These weights are input into the feature fusion layer for multi - scale dilated fusion, which effectively improves the recognition rate of facial expression recognition. In the facial expression recognition task of the 6th ABAW competition, our method achieved excellent results on the official validation set, which fully demonstrates the effectiveness and competitiveness of the proposed method.
This article presents our results for the eighth Affective Behavior Analysis in-the-Wild (ABAW) competition. We combine facial emotional descriptors extracted by pre-trained models, namely, our EmotiEffLib library, with acoustic features and embeddings of texts recognized from speech. The frame-level features are aggregated and fed into simple classifiers, e.g., multi-layered perceptron (feed-forward neural network with one hidden layer), to predict ambivalence/hesitancy and facial expressions. In the latter case, we also use the pre-trained facial expression recognition model to select high-score video frames and prevent their processing with a domain-specific video classifier. The video-level prediction of emotional mimicry intensity is implemented by simply aggregating frame-level features and training a multi-layered perceptron. Experimental results for three tasks from the ABAW challenge demonstrate that our approach significantly increases validation metrics compared to existing baselines.