DisFace3DNet is a facial attractiveness prediction model that disentangles 3D facial components—identity, skin, hair, light, background, expression, and pose—to learn seven component reference scores from overall ratings using weak semantic supervision. The model combines five static and two signed dynamic scores to produce an overall rating, revealing each component’s numerical contribution and enabling component‑specific comparisons across images. On the SCUT‑FBP5500 dataset, it achieves a Pearson correlation of 0.8904±0.0063 with human ratings, with skin, hair, and facial shape contributing most to prediction variation.
arXiv:2607. 16287v1 Announce Type: cross Abstract: Neural Radiance Fields (NeRF) have enabled photorealistic novel-view synthesis of 3D scenes and, in the facial domain, have been extended to reconstruct and animate 3D face models from a small number of images.
Video Diffusion Avatars (VDAs) fine‑tune video diffusion models to generate photo‑realistic, multi‑view consistent avatars from a single image, allowing precise control over camera viewpoint, facial expression, and upper‑body pose. VDAs replace traditional 3D morphable models with a learned expression space, overcoming viewpoint bias through disentanglement and enabling independent control of expression, pose, and viewpoint via proxy‑mesh renderings. The method trains efficiently—34 hours on a single A100—and outperforms state‑of‑the‑art single‑image 3D avatars, improving novel‑view reenactment metrics and producing physically plausible hair dynamics, shadows, reflections, and refractions.
arXiv:2511.05575v2 Announce Type: replace Abstract: Diffusion-based approaches have recently achieved strong results in face swapping, offering improved visual quality over traditional GAN-based meth...
arXiv:2608.22655v1 Announce Type: new Abstract: Creating re-topologized 3D facial meshes is essential for high-quality facial animation but remains labor-intensive and time-consuming. This dissertati...
The paper introduces the 4D Facial Expression Intensity Dataset (4DFEID), comprising 2,869 mesh sequences that capture 3D, temporally continuous facial expressions with varied peak intensities and identities. Subjective intensity ratings were collected via crowdsourcing, yielding over 90,000 Likert-scale annotations. Baseline experiments show that spatial‑temporal graph models outperform traditional frame‑aggregation methods, highlighting the dataset’s value for dynamic 3D expression analysis.
arXiv:2606.10142v2 Announce Type: replace Abstract: Recent advances in 3D generation have led to substantial improvements in realism, controllability, and efficiency, yet the evaluation of 3D assets...
arXiv:2603.06399v2 Announce Type: replace Abstract: Facial attribute classification relies on large-scale annotated datasets in which many traits, such as age and expression, are inherently ambiguous...
The paper introduces a disagreement‑aware dynamic facial expression recognition framework that directly learns from raw annotator vote vectors using a Dirichlet‑Multinomial likelihood, preserving both predictive mean and scale‑sensitive supervision. It adds an ambiguity head to estimate annotation entropy for unseen clips and employs a Chow‑style reject rule that integrates ambiguity, vacuity, temporal instability, and input quality for selective prediction. On the DFEW benchmark, the method maintains recognition accuracy while cutting expected calibration error by 30 % and area‑under‑risk‑curve by 15 %, with predicted ambiguity correlating 0.52 (Spearman) with true annotation entropy, and these gains transfer to FERV39k and hold under identity‑ and movie‑disjoint splits.
Dynamic facial expression recognition (DFER) benchmarks such as DFEW provide multiple annotator votes per clip, yet most models collapse them to a majority label and cannot represent human disagreemen...
Beauty assessments from Multimodal Large Language Models (MLLMs) are increasingly popular, prompting a study comparing 2,513 human ratings to four commercial AI models—Claude, Gemini, GPT, and Grok. The study found that MLLMs consistently rate faces more favorably and with a narrower range than humans, yet they maintain strong correlations with human judgments and accurately track the rank‑ordering of faces. While all models agree strongly with each other, Grok showed the lowest agreement with human ratings, and only face age emerged as a common predictor of attractiveness across humans and MLLMs.
DeepfakeGenome (DFG) is a new benchmark for deepfake detection and attribution that includes 100 facial forgery algorithms and 2 million images, making it 4 to 100 times larger than previous datasets. The benchmark introduces four evaluation protocols, notably a retrieval-based attribution method that aligns with real-world blacklist registration scenarios. Extensive experiments—over 2,000 evaluations—yield ten key findings, offering fresh insights into the deepfake attribution field.
arXiv:2609.13255v1 Announce Type: new Abstract: Facial state analysis plays a crucial role in understanding human expressions, psychological modeling, and human computer interaction. Traditional unim...