Speech-driven 3D facial animation is important for many multimedia applications. Recent work has shown promise in using either Diffusion models or Transformer architectures for this task. However, their mere aggregation does not lead to improved performance. We suspect this is due to a shortage of paired audio-4D data, which is crucial for the Transformer to effectively perform as a denoiser within the Diffusion framework. To tackle this issue, we present DiffSpeaker, a Transformer-based network equipped with novel biased conditional attention modules. These modules serve as substitutes for the traditional self/cross-attention in standard Transformers, incorporating thoughtfully designed biases that steer the attention mechanisms to concentrate on both the relevant task-specific and diffusion-related conditions. We also explore the trade-off between accurate lip synchronization and non-verbal facial expressions within the Diffusion paradigm. Experiments show our model not only achieves state-of-the-art performance on existing benchmarks, but also fast inference speed owing to its ability to generate facial motions in parallel.
Network traffic refers to the amount of information being sent and received over the internet or any system that connects computers. Analyzing and understanding network traffic is vital for improving network security and management. However, the analysis of network traffic poses great challenges due to the unique characteristics of data packets, such as heterogeneous headers and encrypted payload lacking semantics. To capture the latent semantics of traffic, a few studies have adopted pre-training techniques based on the Transformer encoder or decoder to learn the representations from large-scale traffic data. However, these methods typically excel only in traffic understanding (classification) or traffic generation tasks. To address this issue, we develop Lens, a foundational network traffic model that leverages the T5 architecture to learn the pre-trained representations from large-scale unlabeled data. Harnessing the strength of the encoder-decoder framework, which captures the global information while preserving the generative ability, our model can better learn the representations from large-scale network traffic. To further enhance pre-training performance, we design a novel loss that integrates three distinct tasks, namely Masked Span Prediction (MSP), Packet Order Prediction (POP), and Homologous Traffic Prediction (HTP). Evaluation results on multiple benchmark datasets demonstrate that the proposed Lens outperforms the baselines in most downstream tasks related to both traffic understanding and traffic generation. Notably, it also requires considerably less labeled data for fine-tuning compared to current methods.
We introduce a new generative model that combines latent diffusion with persistent homology to create 3D shapes with high diversity, with a special emphasis on their topological characteristics. Our method involves representing 3D shapes as implicit fields, then employing persistent homology to extract topological features, including Betti numbers and persistence diagrams. The shape generation process consists of two steps. Initially, we employ a transformer-based autoencoding module to embed the implicit representation of each 3D shape into a set of latent vectors. Subsequently, we navigate through the learned latent space via a diffusion model. By strategically incorporating topological features into the diffusion process, our generative module is able to produce a richer variety of 3D shapes with different topological structures. Furthermore, our framework is flexible, supporting generation tasks constrained by a variety of inputs, including sparse and partial point clouds, as well as sketches. By modifying the persistence diagrams, we can alter the topology of the shapes generated from these input modalities.
Multi-modal Large Language Models (MLLMs) have shown impressive abilities in generating reasonable responses with respect to multi-modal contents. However, there is still a wide gap between the performance of recent MLLM-based applications and the expectation of the broad public, even though the most powerful OpenAI's GPT-4 and Google's Gemini have been deployed. This paper strives to enhance understanding of the gap through the lens of a qualitative study on the generalizability, trustworthiness, and causal reasoning capabilities of recent proprietary and open-source MLLMs across four modalities: ie, text, code, image, and video, ultimately aiming to improve the transparency of MLLMs. We believe these properties are several representative factors that define the reliability of MLLMs, in supporting various downstream applications. To be specific, we evaluate the closed-source GPT-4 and Gemini and 6 open-source LLMs and MLLMs. Overall we evaluate 230 manually designed cases, where the qualitative results are then summarized into 12 scores (ie, 4 modalities times 3 properties). In total, we uncover 14 empirical findings that are useful to understand the capabilities and limitations of both proprietary and open-source MLLMs, towards more reliable downstream multi-modal applications.
Diffusion models have demonstrated remarkable efficacy in various generative tasks with the predictive prowess of denoising model. Currently, these models employ a uniform denoising approach across all timesteps. However, the inherent variations in noisy latents at each timestep lead to conflicts during training, constraining the potential of diffusion models. To address this challenge, we propose a novel two-stage training strategy termed Step-Adaptive Training. In the initial stage, a base denoising model is trained to encompass all timesteps. Subsequently, we partition the timesteps into distinct groups, fine-tuning the model within each group to achieve specialized denoising capabilities. Recognizing that the difficulties of predicting noise at different timesteps vary, we introduce a diverse model size requirement. We dynamically adjust the model size for each timestep by estimating task difficulty based on its signal-to-noise ratio before fine-tuning. This adjustment is facilitated by a proxy-based structural importance assessment mechanism, enabling precise and efficient pruning of the base denoising model. Our experiments validate the effectiveness of the proposed training strategy, demonstrating an improvement in the FID score on CIFAR10 by over 0.3 while utilizing only 80\% of the computational resources. This innovative approach not only enhances model performance but also significantly reduces computational costs, opening new avenues for the development and application of diffusion models.
Recent advancements in large language models (LLMs) have brought significant changes to various domains, especially through LLM-driven autonomous agents. These agents are now capable of collaborating seamlessly, splitting tasks and enhancing accuracy, thus minimizing the need for human involvement. However, these agents often approach a diverse range of tasks in isolation, without benefiting from past experiences. This isolation can lead to repeated mistakes and inefficient trials in task solving. To this end, this paper introduces Experiential Co-Learning, a novel framework in which instructor and assistant agents gather shortcut-oriented experiences from their historical trajectories and use these past experiences for mutual reasoning. This paradigm, enriched with previous experiences, equips agents to more effectively address unseen tasks.
This paper presents a novel two-stage approach for reconstructing human faces from sparse-view images, a task made challenging by the unique geometry and complex skin reflectance of each individual. Our method focuses on decomposing key facial attributes, including geometry, diffuse reflectance, and specular reflectance, from ambient light. Initially, we create a general facial template from a diverse collection of individual faces, capturing essential geometric and reflectance characteristics. Guided by this template, we refine each specific face model in the second stage, which further considers the interaction between geometry and reflectance, as well as the subsurface scattering effects on facial skin. Our method enables the reconstruction of high-quality facial representations from as few as three images, offering improved geometric accuracy and reflectance detail. Through comprehensive evaluations and comparisons, our method demonstrates superiority over existing techniques. Our method effectively disentangles geometry and reflectance components, leading to enhanced quality in synthesizing new views and opening up possibilities for applications such as relighting and reflectance editing. We will make the code publicly available.
Human-centric perception (e.g. pedetrian detection, segmentation, pose estimation, and attribute analysis) is a long-standing problem for computer vision. This paper introduces a unified and versatile framework (HQNet) for single-stage multi-person multi-task human-centric perception (HCP). Our approach centers on learning a unified human query representation, denoted as Human Query, which captures intricate instance-level features for individual persons and disentangles complex multi-person scenarios. Although different HCP tasks have been well-studied individually, single-stage multi-task learning of HCP tasks has not been fully exploited in the literature due to the absence of a comprehensive benchmark dataset. To address this gap, we propose COCO-UniHuman benchmark dataset to enable model development and comprehensive evaluation. Experimental results demonstrate the proposed method's state-of-the-art performance among multi-task HCP models and its competitive performance compared to task-specific HCP models. Moreover, our experiments underscore Human Query's adaptability to new HCP tasks, thus demonstrating its robust generalization capability. Codes and data will be publicly accessible.
Neural radiance fields (NeRF) typically require a complete set of images taken from multiple camera perspectives to accurately reconstruct geometric details. However, this approach raise significant privacy concerns in the context of facial reconstruction. The critical need for privacy protection often leads invidividuals to be reluctant in sharing their facial images, due to fears of potential misuse or security risks. Addressing these concerns, we propose a method that leverages privacy-preserving images for reconstructing 3D head geometry within the NeRF framework. Our method stands apart from traditional facial reconstruction techniques as it does not depend on RGB information from images containing sensitive facial data. Instead, it effectively generates plausible facial geometry using a series of identity-obscured inputs, thereby protecting facial privacy.