Face Model Research Articles

Finite Element Analysis of the Influence of the Z-Shaped Position of Cleft lip on the Postoperative Appearance.

In traditional surgery for patients with secondary deformities of cleft lip, the skin is usually cut open, and forceps are used according to the position of the flap. However, this method cannot predict the postoperative effect before surgery, especially the changes in the patient's lip. Moreover, in surgery, the design of the surgical flap is also quite random, needing more quantitative standards for its position and size, which makes it impossible to optimize the surgical effect, resulting in the surgical effect relying on the surgeon's surgical level. In this study, we attempted to scan the patient's face using infrared scanning equipment, establish a three-dimensional model of the patient's face, and compare the postoperative effects under different surgical positions through finite element analysis, especially comparing the displacement of the affected labial peak point to find the optimal surgical plan. In this way, the authors can predict the surgical effect before cleft lip and cleft palate surgery so that patients can directly see the effect after surgery and choose the most suitable scheme.

Preliminary application study of digital technology for constructing three-dimensional facial symmetry reference planes in anterior dental esthetic restoration

Objective: To explore the impact of digital technology in constructing a three-dimensional (3D) symmetry reference plane (SRP) for esthetic restoration of anterior teeth and to evaluate its clinical applicability. Methods: A cross-sectional study was conducted from February to May 2024, involving 20 patients [11 males and 9 females, aged (36.8±11.4) years] who underwent anterior esthetic restorations at the Department of Stomatology, People's Hospital of Ningxia Hui Autonomous Region. Symmetrical reference planes of patients' 3D facial models were constructed using three different algorithms: weighted Procrustes analysis (WPA), Procrustes analysis (PA) based on the ontology-mirror correlation method, and iterative closest point (ICP). The SRP defined by an associate chief physician served as the control (true-value group). The angular errors between each algorithm group and the true-value group were compared. The optimal algorithm was selected and combined with a three-dimensional digital smile design (DSD) to create virtual patients, followed by designing anterior restorations. The visual analogue scale (VAS) was used by patients to score the aesthetic restoration results of the conventional design (control group) and the algorithm-based design (algorithm group). Results: The angular errors of the WPA, PA, and ICP groups were 1.43°±0.66°, 1.82°±0.88°, and 4.74°±2.03° respectively, with statistically significant differences among the groups (F=41.10, P<0.001). Pairwise comparisons showed that the WPA group had significantly smaller angular errors compared to the PA and ICP groups (P<0.05). The VAS scores for aesthetic restoration were significantly higher in the algorithm group (8.09±0.74) compared to the control group (6.30±1.38) (t=-5.49, P<0.001). Conclusions: The SRP constructed using the WPA algorithm demonstrated minimal angular error when compared to the expert-defined SRP and is considered the optimal choice in clinical practice, yielding high patient satisfaction.

Automated pipeline for linear and volumetric assessment of facial swelling after third molar surgery

BackgroundExtraction of mandibular third molars (M3Ms) is a routine procedure in oral and maxillofacial surgery, often associated with postoperative symptoms like pain, facial swelling, and trismus. This study aimed to introduce a standardized and automated protocol for swelling analysis following M3M surgery, presenting results regarding clinical conditions immediately and one-week after surgery.MethodsIn a prospective study, 35 patients were enrolled (mean age: 24.4 ± 5.8 years) for removal of 54 M3Ms. Facial swelling was evaluated through 3D facial scans before surgery (T0), at three days (T1), and seven days (T2) post-surgery. The open-source software 3DSlicer facilitated automated analysis, including data anonymization, orientation, surface registration, qualitative comparisons, linear measurements, and volumetric quantification. Pairwise superimposition of facial models enabled qualitative, vectorial, and quantitative assessments, comparing initial conditions with swelling development at T1 and T2. Additionally, changes between T1 and T2 were also evaluated. Secondary outcomes encompassed clinical evaluations of pain, trismus (maximum mouth opening), and surgery time. Statistical analysis involved the paired Student t-test to assess longitudinal changes and analysis of variance to evaluate outcome variables concerning difficulty scores. Linear regression models correlated primary outcome variables with secondary study variables (α < 0.05).ResultsLongitudinal analysis demonstrated significant but variable facial swelling, pain, and trismus at T1, followed by improvement at T2 (p < 0.001). Linear and volumetric differences correlated positively with surgery time (p < 0.05). A direct proportionality between linear and volume differences was observed, higher values at T1 correlated with higher values at T2 (p < 0.05).ConclusionsAn innovative digital workflow for precise quantification of postoperative facial changes was implemented, incorporating volumetric measurements that surpass linear assessments. Clinical conditions demonstrated a direct correlation with surgery time, deteriorating immediately and improving one-week after surgery.

Enhancing the Aesthetics of 3D Shapes via Reference-based Editing

While there have been previous works that explored methods to enhance the aesthetics of images, the automated beautification of 3D shapes has been limited to specific shapes such as 3D face models. In this paper, we introduce a framework to automatically enhance the aesthetics of general 3D shapes. Our approach employs a reference-based beautification strategy. We first performed data collection to gather the aesthetics ratings of various 3D shapes to create a 3D shape aesthetics dataset. Then we perform reference-based editing to edit the input shape and beautify it by making it look more like some reference shape that is aesthetic. Specifically, we propose a reference-guided global deformation framework to coherently deform the input shape such that its structural proportions will be closer to those of the reference shape. We then optionally transplant some local aesthetic parts from the reference to the input to obtain the beautified output shapes. Comparisons show that our reference-guided 3D deformation algorithm outperforms existing techniques. Furthermore, quantitative and qualitative evaluations demonstrate that the performance of our aesthetics enhancement framework is consistent with both human perception and existing 3D shape aesthetics assessment.

Comparison of Different 3D Surface Registration-Based Methods to Assess Facial Asymmetry.

Facial asymmetry is gaining an increasing diagnostic interest in many clinical contexts. Several three-dimensional surface-based methods have been proposed for its assessment; however, they might provide non-equivalent data. Since there is a lack of comparative studies in these terms, this study aims to compare three methods for assessing the asymmetry of the face and facial thirds, thus addressing whether the potential differences can be considered clinically acceptable or not. Two 'maxillofacial' methods based on the trigeminal nerve distribution and one 'orthodontic' method based on reference horizontal planes were used to identify the facial thirds on 3D facial models of 80 Italian healthy adults to calculate the asymmetry of the face, and the upper, middle, and lower thirds of the face differently selected by each method. As a measure of asymmetry, the Root Mean Square value was calculated through a mirroring surface-based registration. Intra- and inter-operator reliability was verified for each method. Differences and interchangeability between the methods were tested, respectively, by two-way repeated measures ANOVA (Analysis of Variance) and Bland-Altman and Similarity Percentage model analysis. Additionally, the time required to perform each method was assessed. All methods demonstrated excellent intra- and inter-operator reliability. While the ANOVA analysis found significant differences (p < 0.001) for the majority of facial Regions of Interest between each method, the Bland-Altman analysis revealed that the differences were clinically acceptable (<0.50 mm) for all facial regions between the trigeminal methods, and for the face and the upper third of the face between the orthodontic method, which was revealed to be faster, and the trigeminal ones. The additional similarity percentage model provided visual support for the complete interchangeability of the two trigeminal methods, as evidenced by the lower Coefficient of Variation value. There is no best method for assessing facial asymmetry that applies to all types of clinical settings, as we have shown that different methods may not be completely interchangeable. However, we suggest that the methods based on the trigeminal subdivision can be used interchangeably in contexts where the morpho-functional analysis of maxillofacial regions with different embryological origins is considered. Thus, the clinical setting imposes the choice of one method over another and, as we have pointed out, the consequent comparison of data with those obtained with methods whose interchangeability has been demonstrated.

Face Modeling Based on Deep Learning and Traditional Methods: A Survey

Abstract. 3D face modeling, as an important research field in computer graphics and vision, has undergone remarkable development in recent years. The technology is widely used in various scenarios, like film and television production, virtual reality, augmented reality, facial recognition, security monitoring, and medical diagnosis. This paper aims to systematically explore the main technologies of 3D face modeling, especially the application and development of deep learning methods in it, review traditional face modeling methods, introduce in detail the latest technical advances based on deep learning, and analyze the advantages and disadvantages of these methods in practical applications.

Personalized Fitting of Respiratory Mask Using Deep Learning and 3D Facial Modeling

Personalized Fitting of Respiratory Mask Using Deep Learning and 3D Facial Modeling

Accuracy Assessment of EM3D App-Based 3D Facial Scanning Compared to Cone Beam Computed Tomography.

Background: The use of 3D facial scans is becoming essential for dental practice. However, traditional scanners require labor-intensive procedures and are expensive, making them less accessible in routine clinical practice. In this context, high-performance smartphones and dedicated apps offer a more accessible alternative. This study aims to validate the accuracy of the EM3D app, which utilizes the iPhone's TrueDepth camera technology, by comparing it to Cone Beam Computed Tomography (CBCT). Methods: Thirty patients requiring CBCT scans were recruited for the study. Facial scans obtained with the TrueDepth camera of the iPhone 13 Pro in conjunction with EM3D app were automatically superimposed onto the 3D models derived from the CBCTs through the implementation of a deep learning methodology. The approach enabled the automatic identification of fifteen landmarks to perform linear and angular measurements for quantitative assessment. A color map was created to highlight discrepancies between the overlaid meshes, and the overall surface differences between the models were automatically quantified. Results: The overall surface difference between the CBCT and EM3D scans was highly accurate, with a mean discrepancy of 0.387 ± 0.361 mm. The mean discrepancies of most measurements were lower than 1 mm (five out of six; 83.33%) between the groups, with no significant differences (p > 0.05). Conclusions: The combination of the iPhone's TrueDepth camera and the EM3D app exhibited high accuracy for 3D facial modeling. This makes it a cost-effective alternative to professional scanning systems.

Evaluation of the effect of orthodontic treatment on the reliability of facial recognition by using three-dimensional model superimposition technique

3D-3D registration of facial models, has great advantages in personal individual identification in forensic medicine. However, orthodontic treatment has brought changes in facial soft and hard tissues, which has a potential effect in personal identification. The aim of the study was to explore whether orthodontic treatment affects 3D-3D facial recognition. A total of 68 patients aged between 18 and 38 were selected (30 subjects with tooth extraction, 38 subjects with non-tooth extraction) and a control group consisting of 30 volunteers without orthodontic treatment was selected from a database of 3dMD facial models. 3dMD facial model acquisition was performed twice for each subject. For both extraction and non-extraction groups, T0 was acquired before treatment, and T1 was acquired at the end of treatment. The time span of 2 times 3dMD facial model acquisition in the control group was about 2 years. 3dMD facial models were then registered onto other 3dMD models belonging to the same and different individuals according to the minimum point-to-point distance, getting a cohort of matches and mismatches. Root mean square (RMS) value of the minimum point-to-point distance between two models was then calculated. The intra- and inter-observer repeatability coefficients were 0.986 (P<0.001) and 0.982 (P<0.001). The absolute technical error of measurement (TEM) value was 0.05 mm and 0.04 mm, and relative technical error of measurement (rTEM) value was 4.91 % and 3.60 %, respectively. Possible significant differences between groups were assessed through independent samples t-test or Mann–Whitney U test (p<0.001). The average RMS value was 1.03±0.33 mm in matches of orthodontic group, 0.67±0.22 mm in matches of control group, and 2.66±0.51 mm in mismatches. An RMS value range of 1.70–1.77 mm could distinguish matches from mismatches in 100 % of cases in the present study. This study showed that orthodontic treatment would not affect 3D-3D facial recognition in adults for the purpose of individual identification in forensic medicine.

Analysis of friction vibration characteristics of the end face of a T-slot sealing ring under thermal deformation

Aiming at the friction vibration problem caused by dry friction contact during the start-stop process of a dry gas sealing device, this paper takes the end face of a T-slot dry gas sealing ring as the research object and establishes a numerical calculation model of the end face of the T-slot sealing ring. The thermal coupling method is used to obtain the thermal deformation law of the seal ring end face. The transient dynamics analysis method was applied to study the frequency and vibration mode of the friction vibration instability of the seal ring end face under thermal deformation. A vibration test bench was set up for the friction ring, and the friction vibration signals of the test pieces were collected, and the characteristic frequencies were extracted. The results show that: After thermal deformation, the end face of the sealing ring changes from a parallel surface to an inward converging surface, and the friction system of the sealing ring will become unstable. When considering the influence of thermal deformation, there are more unstable frequencies and lower frequency values that are most prone to frictional vibration. As the thermal deformation of the sealing ring increases, the frictional vibration becomes greater. In areas where the characteristic frequency values of frictional vibration are concentrated during thermal deformation, the occurrence rate of frictional vibration noise is higher.

Evaluation of Benzene Presence and Formation in Benzoyl Peroxide Drug Products

The potent carcinogen, benzene, is a known degradation product of benzoyl peroxide (BPO) and was recently reported to form when BPO drug products, used for acne and rosacea treatment, are incubated at body temperature and elevated temperatures expected during storage and transportation. This study provides evidence for a wide range of benzene concentrations (0.16 ppm to 35.30 ppm) detected by GC-MS in 111 over-the-counter BPO drug products tested and maintained at room temperature. A prescription encapsulated BPO drug product was stability tested at cold (2°C) and elevated temperature (50°C), resulting in no apparent benzene formation at 2°C, and high levels of benzene formation at 50°C, suggesting that encapsulation technology may not stabilize BPO drug products but cold storage may greatly reduce benzene formation. Face model experiments where BPO drug product was applied to PolyMethyl MethAcrylate (PMMA) photoprotection test skin plates and benzene was detected in surrounding air by SIFT-MS, showed detectable benzene through evaporation and substantial benzene formation when exposed to UV light at levels below peak sunlight. Results suggest that potential benzene exposure from formation during BPO drug product use poses significant risks independent of the starting benzene concentration.

ANTI-AGING COSMETIC PROCEDURES: BENEFITS AND RISKS

The rising demand for aesthetic interventions reflects a broader cultural trend toward enhancing physical appearance and rejuvenation, fueled by technological advancements and increased accessibility of invasive and non-invasive cosmetic procedures. Skin aging, driven by a combination of genetic, hormonal, and environmental factors, results in visible changes such as wrinkles and reduced skin elasticity. The growing availability of cosmetic procedures, including botulinum toxin injections, dermal fillers, and surgical options like rhinoplasty and liposuction, underscores a shift in public attitudes toward cosmetic enhancement and concealing of visible signs of aging. Social media platforms play a crucial role in influencing treatment decisions, changing perceptions and practices of cosmetic procedures. However, uncontrolled expansion of these procedures is associated with risks, including psychological distress and physical complications. Technological innovations, such as ultrasound-guided injections and 3D facial modelling, are improving procedural accuracy, patient safety, and satisfaction. This review examines the driving factors behind the surge in cosmetic procedures, addresses some issues in the field, and evaluates how emerging advanced technologies are enhancing outcomes and safety.

Multimodal Fusion for Talking Face Generation Utilizing Speech-Related Facial Action Units

Talking face generation is to synthesize a lip-synchronized talking face video by inputting an arbitrary face image and corresponding audio clips. The current talking face model can be divided into four parts: visual feature extraction, audio feature processing, multimodal feature fusion, and rendering module. For the visual feature extraction part, existing methods face the challenge of complex learning task with noisy features, this article introduces an attention-based disentanglement module to disentangle the face into Audio-face and Identity-face using speech-related facial action unit (AU) information. For the multimodal feature fusion part, existing methods ignore not only the interaction and relationship of cross-modal information but also the local driving information of the mouth muscles. This study proposes a novel generative framework that incorporates a dilated non-causal temporal convolutional self-attention network as a multimodal fusion module to enhance the learning of cross-modal features. The proposed method employs both audio- and speech-related facial AUs as driving information. Speech-related AU information can facilitate more accurate mouth movements. Given the high correlation between speech and speech-related AUs, we propose an audio-to-AU module to predict speech-related AU information. Finally, we present a diffusion model for the synthesis of talking face images. We verify the effectiveness of the proposed model on the GRID and TCD-TIMIT datasets. An ablation study is also conducted to verify the contribution of each component. The results of quantitative and qualitative experiments demonstrate that our method outperforms existing methods in terms of both image quality and lip-sync accuracy. Code is available at https://mftfg-au.github.io/Multimodal_Fusion/.

Weakly Supervised 3D Face Reconstruction with Joint Spatial and Frequency Domain Information

3D face reconstruction is an important research direction in computer vision, and its goal is to recover a 3D face model from a single face picture. In the absence of real 3D face data, how to reconstruct a 3D face with a high degree of realism has become a hot research topic in recent years. Existing reconstruction algorithms usually rely on 3D labels generated from a large number of 2D face images as training data, however, inaccurate data will seriously affect the reconstruction quality. For this reason, the paper proposes a joint spatial-frequency domain decoupled weak supervision to achieve 3D face reconstruction, the main idea of which is to construct a multi-level loss function by using the weakly supervised information extracted from the spatial domain, and separating the frequency-domain information between the input and the rendered image in the frequency domain, and minimizing the difference between the two by difference computation. The method combines deep learning with 3D deformable models to reconstruct 3D models with high quality texture and shape from only a single face image. Quantitative experiments on the AFLW2000-3D and MICC Florence datasets show that the normalized average error in the small pose interval is as low as 2.42%, and the face reconstruction accuracy in the outdoor scene is 0.98 0.22 mm. Qualitative experiments on the MoFa-test, MICA datasets show that when faced with reconstruction with different poses, lighting, and expressions , our method outperforms other state-of-the-art reconstruction methods.

FaReT 2.1: Anatomically precise manipulation of race in 3D face models and a pipeline to import real face scans

FaReT 2.1: Anatomically precise manipulation of race in 3D face models and a pipeline to import real face scans

AnyFace++: Deep Multi-Task, Multi-Domain Learning for Efficient Face AI.

Accurate face detection and subsequent localization of facial landmarks are mandatory steps in many computer vision applications, such as emotion recognition, age estimation, and gender identification. Thanks to advancements in deep learning, numerous facial applications have been developed for human faces. However, most have to employ multiple models to accomplish several tasks simultaneously. As a result, they require more memory usage and increased inference time. Also, less attention is paid to other domains, such as animals and cartoon characters. To address these challenges, we propose an input-agnostic face model, AnyFace++, to perform multiple face-related tasks concurrently. The tasks are face detection and prediction of facial landmarks for human, animal, and cartoon faces, including age estimation, gender classification, and emotion recognition for human faces. We trained the model using deep multi-task, multi-domain learning with a heterogeneous cost function. The experimental results demonstrate that AnyFace++ generates outcomes comparable to cutting-edge models designed for specific domains.

Face shape transfer via semantic warping

Face reshaping aims to adjust the shape of a face in a portrait image to make the face aesthetically beautiful, which has many potential applications. Existing methods 1) operate on the pre-defined facial landmarks, leading to artifacts and distortions due to the limited number of landmarks, 2) synthesize new faces based on segmentation masks or sketches, causing generated faces to look dissatisfied due to the losses of skin details and difficulties in dealing with hair and background blurring, and 3) project the positions of the deformed feature points from the 3D face model to the 2D image, making the results unrealistic because of the misalignment between feature points. In this paper, we propose a novel method named face shape transfer (FST) via semantic warping, which can transfer both the overall face and individual components (e.g., eyes, nose, and mouth) of a reference image to the source image. To achieve controllability at the component level, we introduce five encoding networks, which are designed to learn feature embedding specific to different face components. To effectively exploit the features obtained from semantic parsing maps at different scales, we employ a straightforward method of directly connecting all layers within the global dense network. This direct connection facilitates maximum information flow between layers, efficiently utilizing diverse scale semantic parsing information. To avoid deformation artifacts, we introduce a spatial transformer network, allowing the network to handle different types of semantic warping effectively. To facilitate extensive evaluation, we construct a large-scale high-resolution face dataset, which contains 14,000 images with a resolution of 1024 × 1024. Superior performance of our method is demonstrated by qualitative and quantitative experiments on the benchmark dataset.

Driving mutual advancement of 3D reconstruction and inpainting for masked faces

Target occlusion or pollution has always been a common and difficult problem in 3D reconstruction, seriously affecting the reconstruction effect, especially in single image scenario. To address the issues of incomplete reconstruction caused by pixel missing, a novel framework promoting mutual enhancement between 3D reconstruction and inpainting is proposed, which is capable of reconstructing a realistic 3D face with pixel completion from single image without prior assumptions. The framework is composed of an inpainting module for facial pixel completion and a reconstruction module for 3D face modeling. The inpainting module provides complete texture and accurate 3D geometry inferring premise for 3D reconstruction, and the reconstruction module is also involved in face inpainting, constraining the inpainting module to fill plausible pixels that are closer to the ground truth. Experiments show that the accurate 3D faces with complete and fine texture can be reconstructed by the proposed framework from largely masked images, with a competitive performance even surpassing most of state-of-the-art methods with unmasked images as input. Furthermore, this framework has achieved state-of-the-art performance in image inpainting on multiple face datasets.

3DFaceShop: Explicitly Controllable 3D-Aware Portrait Generation.

In contrast to the traditional avatar creation pipeline which is a costly process, contemporary generative approaches directly learn the data distribution from photographs. While plenty of works extend unconditional generative models and achieve some levels of controllability, it is still challenging to ensure multi-view consistency, especially in large poses. In this work, we propose a network that generates 3D-aware portraits while being controllable according to semantic parameters regarding pose, identity, expression and illumination. Our network uses neural scene representation to model 3D-aware portraits, whose generation is guided by a parametric face model that supports explicit control. While the latent disentanglement can be further enhanced by contrasting images with partially different attributes, there still exists noticeable inconsistency in non-face areas when animating expressions. We solve this by proposing a volume blending strategy in which we form a composite output by blending dynamic and static areas, with two parts segmented from the jointly learned semantic field. Our method outperforms prior arts in extensive experiments, producing realistic portraits with vivid expression in natural lighting when viewed from free viewpoints. It also demonstrates generalization ability to real images as well as out-of-domain data, showing great promise in real applications.

Is emotion perception altered by gaze direction, gender appearance, and gender identity of the perceived face?

The purpose of the present study was to examine how gaze and emotion processing may change due to differences in gender appearance and gender identity of the perceived face. We manipulated gender appearance (male or female), gender identity (cisgender or transgender), gaze direction (direct or averted), and expressed emotions (anger, fear, or neutral) of face models in an emotion rating task. We replicate several previous findings, including a direct gaze advantage, an emotion effect, and an interaction between gaze direction and expressed emotion. In line with previous findings on the influence of facial morphology for face processing, we found that male faces were more quickly and intensely perceived for displays of anger, while female faces were more quickly and intensely perceived for displays of fear. Of key interest, gender identity influenced face perception for different emotion expressions and gaze directions for ratings and reaction times in a variety of ways. For example, transgender male faces were seen as angrier and less fearful than cisgender male faces, while the opposite effect occurred for female faces. These results suggest that face perception is systematically shaped by morphological differences as well as more abstract social constructs related to gender identity. (PsycInfo Database Record (c) 2024 APA, all rights reserved).