Facial beauty prediction is an emerging topic. The pursuit of facial beauty is the nature of human beings. As the demand for aesthetic surgery has increased significantly over the past few years, an understanding beauty is becoming increasingly important in medical settings. This work proposes a new ensemble based on the pre-trained convolutional neural network (CNN) models to identify scores for facial beauty prediction. These ensembles were originally built from the following previously trained models: DenseNet-201, Inception-v3, MobileNetV2, and EfficientNetB7. According to the SCUT-FBP5500 benchmark dataset, the proposed model obtains a Pearson coefficient of 0.9469. This reveals that the suggested EN-CNNs model can be successfully applied in a variety of face-to-face applications.
The topic of facial beauty analysis has emerged as a crucial and fascinating subject of human culture. With various applications and significant attention from researchers, recent studies have investigated the relationship between facial features and age, emotions, and other factors using multidisciplinary approaches. Facial beauty prediction is a significant visual recognition problem in the assessment of facial attractiveness, which is consistent with human perception. Overcoming the challenges associated with facial beauty prediction requires considerable effort due to the field’s novelty and lack of resources. In this vein, a deep learning method has recently demonstrated remarkable abilities in feature representation and analysis. Accordingly, this paper proposes an ensemble based on pre-trained convolutional neural network models to identify scores for facial beauty prediction. These ensembles are three separate deep convolutional neural networks, each with a unique structural representation built by previously trained models from Inceptionv3, Mobilenetv2, and a new simple network based on Convolutional Neural Networks (CNNs) for facial beauty prediction problems. According to the SCUT-FBP5500 benchmark dataset, the obtained 0.9350 Pearson coefficient experimental result demonstrated that using this ensemble of deep networks leads to a better prediction of facial beauty closer to human evaluation than conventional technology that spreads facial beauty. Finally, potential research directions are suggested for future research on facial beauty prediction.
Facial beauty prediction is a complex task that involves quantifying human facial attractiveness based on visual features. In this paper, we propose a novel approach utilizing the MobileViT architecture, a hybrid model that integrates convolutional neural networks (CNNs) and vision transformers (ViTs) to tackle this challenge. The MobileViT model is specifically designed to capture both local details and global context, making it highly effective for facial image analysis while ensuring computational efficiency. We evaluated the performance of our model on the SCUT-FBP5500 dataset, a large-scale facial beauty dataset comprising 5,500 images labeled with beauty scores. Our approach achieves state-of-the-art results, demonstrating a Pearson correlation of 0.9515. These results significantly surpass those of existing methods, including CNN-based and semi-supervised approaches. The lightweight and efficient design of MobileViT renders it suitable for real-time applications on mobile and edge devices, highlighting its potential for practical use in beauty prediction tasks.
27 August 2025 in Engineering Applications of Artificial Intelligence
Facial beauty analysis is a crucial subject in human culture among researchers across various applications. Recent studies have utilized multidisciplinary approaches to examine the relationship between facial traits, age, emotions, and other factors. Facial beauty prediction is a significant visual recognition challenge that evaluates facial attractiveness for human perception. This task demands considerable effort due to the novelty of the field and the limited resources available, including a small database for facial beauty prediction. In this context, a deep learning method has recently shown remarkable capabilities in predicting facial beauty. Additionally, vision Transformers have recently been introduced as novel deep learning approaches and have shown strong performance in various applications. The key issue is that the vision transformer performs significantly worse than ResNet when trained on a small ImageNet database. In this paper, we propose to address the challenges of predicting facial beauty by utilizing vision transformers instead of relying on feature extraction based on Convolutional Neural Networks, which are commonly used in traditional methods. Moreover, we define and optimize a set of hyperparameters according to the SCUT-FBP5500 benchmark dataset. The model achieves a Pearson coefficient of 0.9534. Experimental results indicated that using this proposed network leads to better predicting facial beauty closer to human evaluation than conventional technology that provides facial beauty assessment.
Facial beauty prediction (FBP) has garnered significant attention in recent years due to its wide range of applications in cosmetics, entertainment, and healthcare. Traditional approaches rely heavily on handcrafted features or deep convolutional neural networks (CNNs), which often miss the global contextual information crucial for accurate prediction. This paper presents a novel method for facial beauty prediction using a Global Context Vision Transformer (GCViT). By leveraging the Vision Transformer's ability to model long-range dependencies and global relationships between facial features, our approach enhances the prediction accuracy beyond the limitations of CNN-based models. The proposed model integrates CNN-based feature extraction and Transformer-based attention mechanisms to effectively model intricate facial details and global context. A hierarchical architecture is used to progressively refine features across multiple resolutions, enabling the model to focus on fine-grained facial attributes while maintaining global consistency. We evaluate the performance of our model on the SCUT-FBP5500 dataset, achieving state-of-the-art results 0.9456 with significantly higher Pearson Correlation compared to existing methods. The experimental results demonstrate that the GC ViT outperforms traditional CNN-based approaches, offering a more effective solution for facial beauty prediction tasks.