Talking face generation has historically struggled to produce head movements and natural facial expressions without guidance from additional reference videos. Recent developments in diffusion-based generative models allow for more realistic and stable data synthesis and their performance on image and video generation has surpassed that of other generative models. In this work, we present an autoregressive diffusion model that requires only one identity image and audio sequence to generate a video of a realistic talking human head. Our solution is capable of hallucinating head movements, facial expressions, such as blinks, and preserving a given background. We evaluate our model on two different datasets, achieving state-of-the-art results on both of them.
Neuroscience studies have revealed that the brain encodes visual content and embeds information in neural activity. Recently, deep learning techniques have facilitated attempts to address visual reconstructions by mapping brain activity to image stimuli using generative adversarial networks (GANs). However, none of these studies have considered the semantic meaning of latent code in image space. Omitting semantic information could potentially limit the performance. In this study, we propose a new framework to reconstruct facial images from functional Magnetic Resonance Imaging (fMRI) data. With this framework, the GAN inversion is first applied to train an image encoder to extract latent codes in image space, which are then bridged to fMRI data using linear transformation. Following the attributes identified from fMRI data using an attribute classifier, the direction in which to manipulate attributes is decided and the attribute manipulator adjusts the latent code to improve the consistency between the seen image and the reconstructed image. Our experimental results suggest that the proposed framework accomplishes two goals: (1) reconstructing clear facial images from fMRI data and (2) maintaining the consistency of semantic characteristics.
Recent studies show that depression can be partially reflected from human facial attributes. Since facial attributes have various data structure and carry different information, existing approaches fail to specifically consider the optimal way to extract depression-related features from each of them, as well as investigates the best fusion strategy. In this paper, we propose to extend Neural Architecture Search (NAS) technique for designing an optimal model for multiple facial attributes-based depression recognition, which can be efficiently and robustly implemented in a small dataset. Our approach first conducts a warmer up step to the feature extractor of each facial attribute, aiming to largely reduce the search space and providing customized architecture, where each feature extractor can be either a Convolution Neural Networks (CNN) or Graph Neural Networks (GNN). Then, we conduct an end-to-end architecture search for all feature extractors and the fusion network, allowing the complementary depression cues to be optimally combined with less redundancy. The experimental results on AVEC 2016 dataset show that the model explored by our approach achieves breakthrough performance with 27\% and 30\% RMSE and MAE improvements over the existing state-of-the-art. In light of these findings, this paper provides solid evidences and a strong baseline for applying NAS to time-series data-based mental health analysis.
Understanding of human visual perception has historically inspired the design of computer vision architectures. As an example, perception occurs at different scales both spatially and temporally, suggesting that the extraction of salient visual information may be made more effective by paying attention to specific features at varying scales. Visual changes in the body due to physiological processes also occur at different scales and with modality-specific characteristic properties. Inspired by this, we present BigSmall, an efficient architecture for physiological and behavioral measurement. We present the first joint camera-based facial action, cardiac, and pulmonary measurement model. We propose a multi-branch network with wrapping temporal shift modules that yields both accuracy and efficiency gains. We observe that fusing low-level features leads to suboptimal performance, but that fusing high level features enables efficiency gains with negligible loss in accuracy. Experimental results demonstrate that BigSmall significantly reduces the computational costs. Furthermore, compared to existing task-specific models, BigSmall achieves comparable or better results on multiple physiological measurement tasks simultaneously with a unified model.
The rising performance of deep neural networks is often empirically attributed to an increase in the available computational power, which allows complex models to be trained upon large amounts of annotated data. However, increased model complexity leads to costly deployment of modern neural networks, while gathering such amounts of data requires huge costs to avoid label noise. In this work, we study the ability of compression methods to tackle both of these problems at once. We hypothesize that quantization-aware training, by restricting the expressivity of neural networks, behaves as a regularization. Thus, it may help fighting overfitting on noisy data while also allowing for the compression of the model at inference. We first validate this claim on a controlled test with manually introduced label noise. Furthermore, we also test the proposed method on Facial Action Unit detection, where labels are typically noisy due to the subtlety of the task. In all cases, our results suggests that quantization significantly improve the results compared with existing baselines, regularization as well as other compression methods.
Facial expression recognition plays an important role in human-computer interaction. In this paper, we propose the Coarse-to-Fine Cascaded network with Smooth Predicting (CFC-SP) to improve the performance of facial expression recognition. CFC-SP contains two core components, namely Coarse-to-Fine Cascaded networks (CFC) and Smooth Predicting (SP). For CFC, it first groups several similar emotions to form a rough category, and then employs a network to conduct a coarse but accurate classification. Later, an additional network for these grouped emotions is further used to obtain fine-grained predictions. For SP, it improves the recognition capability of the model by capturing both universal and unique expression features. To be specific, the universal features denote the general characteristic of facial emotions within a period and the unique features denote the specific characteristic at this moment. Experiments on Aff-Wild2 show the effectiveness of the proposed CFSP.
In this paper, we propose a new feature descriptor Cross-Centroid Ripple Pattern (CRIP) for facial expression recognition. CRIP encodes the transitional pattern of a facial expression by incorporating cross-centroid relationship between two ripples located at radius r1 and r2 respectively. These ripples are generated by dividing the local neighborhood region into subregions. Thus, CRIP has ability to preserve macro and micro structural variations in an extensive region, which enables it to deal with side views and spontaneous expressions. Furthermore, gradient information between cross centroid ripples provides strenght to captures prominent edge features in active patches: eyes, nose and mouth, that define the disparities between different facial expressions. Cross centroid information also provides robustness to irregular illumination. Moreover, CRIP utilizes the averaging behavior of pixels at subregions that yields robustness to deal with noisy conditions. The performance of proposed descriptor is evaluated on seven comprehensive expression datasets consisting of challenging conditions such as age, pose, ethnicity and illumination variations. The experimental results show that our descriptor consistently achieved better accuracy rate as compared to existing state-of-art approaches.
Talking face generation, also known as speech-to-lip generation, reconstructs facial motions concerning lips given coherent speech input. The previous studies revealed the importance of lip-speech synchronization and visual quality. Despite much progress, they hardly focus on the content of lip movements i.e., the visual intelligibility of the spoken words, which is an important aspect of generation quality. To address the problem, we propose using a lip-reading expert to improve the intelligibility of the generated lip regions by penalizing the incorrect generation results. Moreover, to compensate for data scarcity, we train the lip-reading expert in an audio-visual self-supervised manner. With a lip-reading expert, we propose a novel contrastive learning to enhance lip-speech synchronization, and a transformer to encode audio synchronically with video, while considering global temporal dependency of audio. For evaluation, we propose a new strategy with two different lip-reading experts to measure intelligibility of the generated videos. Rigorous experiments show that our proposal is superior to other State-of-the-art (SOTA) methods, such as Wav2Lip, in reading intelligibility i.e., over 38% Word Error Rate (WER) on LRS2 dataset and 27.8% accuracy on LRW dataset. We also achieve the SOTA performance in lip-speech synchronization and comparable performances in visual quality.
Automated facial age estimation has diverse real-world applications in multimedia analysis, e.g., video surveillance, and human-computer interaction. However, due to the randomness and ambiguity of the aging process, age assessment is challenging. Most research work over the topic regards the task as one of age regression, classification, and ranking problems, and cannot well leverage age distribution in representing labels with age ambiguity. In this work, we propose a simple yet effective loss function for robust facial age estimation via distribution learning, i.e., adaptive mean-residue loss, in which, the mean loss penalizes the difference between the estimated age distribution's mean and the ground-truth age, whereas the residue loss penalizes the entropy of age probability out of dynamic top-K in the distribution. Experimental results in the datasets FG-NET and CLAP2016 have validated the effectiveness of the proposed loss. Our code is available at https://github.com/jacobzhaoziyuan/AMR-Loss.
We propose a real time deep learning framework for video-based facial expression capture. Our process uses a high-end facial capture pipeline based on FACEGOOD to capture facial expression. We train a convolutional neural network to produce high-quality continuous blendshape weight output from video training. Since this facial capture is fully automated, our system can drastically reduce the amount of labor involved in the development of modern narrative-driven video games or films involving realistic digital doubles of actors and potentially hours of animated dialogue per character. We demonstrate compelling animation inference in challenging areas such as eyes and lips.