Layout-to-Image generation has significantly advanced content creation by enabling the rendering of visual text under predefined spatial layouts. Current approaches achieve training-free layout guidance by constructing attention-based energy functions to derive correction gradients. In this paper, we demonstrate that vanilla energy functions suffer from two limitations, resulting in imprecise layout control and visually unrealistic artifacts. First, the normalizing factor of the Boltzmann distribution defined by the energy functions is non-negligible when calculating correction gradients, yet current energy functions cannot compute this factor exactly. Furthermore, while attention varies over time during the denoising process, existing approaches employ a fixed formulation. To address these challenges, we introduce FreLay, a novel training-free approach equipped with a frequency-aware energy function. Our method first reformulates the energy function to handle the normalization factor, enabling accurate computation of correction gradients. Simultaneously, leveraging the prior knowledge that low-frequency information deteriorates slower during noise addition, we design a time-specific energy function for each timestep from a frequency-domain perspective. Experimental results demonstrate that FreLay consistently outperforms existing state-of-the-art training-free methods by a large margin both qualitatively and quantitatively across multiple datasets.

This paper focuses on face stylization with a single artistic target. Existing works for this task often fail to retain the source content while achieving geometry variation. Here, we present a novel StyO model, i.e., Stylize the face in only One-shot, to solve the above problem. In particular, StyO exploits a disentanglement and recombination strategy. It first disentangles the content and style of source and target images into identifiers, which are then recombined in a cross manner to derive the stylized face image. In this way, StyO decomposes complex images into independent and specific attributes, and simplifies one-shot face stylization as the combination of different attributes from input images, thus producing results better matching face geometry of target image and content of source one. StyO is implemented with latent diffusion models (LDM) and composed of two key modules: 1) Identifier Disentanglement Learner (IDL) for disentanglement phase. It represents identifiers as contrastive text prompts, i.e. positive and negative descriptions. And it introduces a novel triple reconstruction loss to fine-tune the pre-trained LDM for encoding style and content into corresponding identifiers; 2) Fine-graind Content Controller (FCC) for recombination phase. It recombines disentangled identifiers from IDL to form an augmented text prompt for generating stylized faces. In addition, FCC also constrains the cross-attention maps of latent and text features to preserve source face details in results. The extensive evaluation shows that StyO produces high-quality images on numerous paintings of various styles and outperforms the current state-of-the-art.

Generating dense multiview images from text prompts is crucial for creating high-fidelity 3D assets. Nevertheless, existing methods struggle with space-view correspondences, resulting in sparse and low-quality outputs. In this paper, we introduce CoSER, a novel consistent dense Multiview Text-To-Image Generator for Text-To-3D, achieving both efficiency and quality by meticulously learning neighbor-view coherence and further alleviating ambiguity through the swift traversal of all views. For achieving neighbor-view consistency, each viewpoint densely interacts with adjacent viewpoints to perceive the global spatial structure, and aggregates information along motion paths explicitly defined by physical principles to refine details. To further enhance cross-view consistency and alleviate content drift, CoSER rapidly scan all views in spiral bidirectional manner to aware holistic information and then scores each point based on semantic material. Subsequently, we conduct weighted down-sampling along the spatial dimension based on scores, thereby facilitating prominent information fusion across all views with lightweight computation. Technically, the core module is built by integrating the attention mechanism with a selective state space model, exploiting the robust learning capabilities of the former and the low overhead of the latter. Extensive evaluation shows that CoSER is capable of producing dense, high-fidelity, content-consistent multiview images that can be flexibly integrated into various 3D generation models.

Local features aggregation and global information perception are the fundamental to point cloud segmentation. However, existing works often fall short in effectively identifying semantic relevant neighbors and face challenges in endowing each point with high-level information. Here, we propose CamPoint, an innovative method that employs virtual cameras to solve the above problems. The core of CamPoint lies in introducing the novel camera visibility feature for points, where each dimension encodes the visibility of that point from a specific camera. Leveraging this feature, we propose the camera perspective slice distance for accurate relevant neighbor searching and design the camera parameter embedding to deliver rich feature representations for global interaction. Specifically, the camera perspective slice distance between two points is defined as a similarity metric derived from their camera visibility features, whereby an increased number of shared cameras observing both points corresponds to a reduced distance between them. To effectively facilitate global semantic perception, we assign each camera an optimizable embedding and then integrate these embeddings into the original spatial features based on visibility attributes, thereby obtaining high-level features enriched with camera priors. Additionally, the state space model characterized by linear computational complexity is employed as the operator to achieve global learning with efficiency. Comprehensive experiments on multiple datasets show that our CamPoint surpasses the current state-of-the-art in multiple datasets, achieving low training cost and fast inference speed.

Recent diffusion models have made significant advancements in generating lifelike videos from driving signals, including a reference character and a skeleton sequence. Nevertheless, these models often struggle with maintaining fidelity, as the generated results frequently deviate in character features, e.g., appearance and identity from the reference. We attribute this issue to the use of driving signals from the same individual during the training process, which biases the model towards skeleton-based shape features and limits its capacity to fully exploit character-specific information, and propose DreamHA to address this issue. DreamHA incorporates diffusion models with Rigid Transformation Augmentation (RTAug), a simple yet effective technique to perturb the shape characteristics of training data, improving the capability of diffusion models in capturing basic appearance features. Additionally, we introduce Identity Keeper (IK) to provide fine-grained facial control and enhance identity consistency. Extensive experimental results demonstrate that our method outperforms state-of-the-art approaches, producing more faithful and consistent animations.

Solving the Nash equilibrium in normal-form games with large-scale strategy spaces presents significant challenges. Open-ended learning frameworks, such as PSRO and its variants, have emerged as effective solutions. However, these methods often lack an efficient metric for evaluating strategy improvement, which limits their effectiveness in approximating equilibria. In this paper, we introduce a novel evaluative metric called Advantage, which possesses desirable properties inherently connected to the Nash equilibrium, ensuring that each strategy update approaches equilibrium. Building upon this, we propose the Advantage Policy Space Response Oracle (A-PSRO), an innovative unified open-ended learning framework applicable to both zero-sum and general-sum games. A-PSRO leverages the Advantage as a refined evaluation metric, leading to a consistent learning objective for agents in normal-form games. Experiments showcase that A-PSRO significantly reduces exploitability in zero-sum games and improves rewards in general-sum games, outperforming existing algorithms and validating its practical effectiveness.

Layout-to-Image generation aims to create complex scenes with precise control over the placement and arrangement of subjects. Existing works have demonstrated that pre-trained Text-to-Image diffusion models can achieve this goal without training on any specific data; however, they often face challenges with imprecise localization and unrealistic artifacts. Focusing on these drawbacks, we propose a novel training-free method, WinWinLay. At its core, WinWinLay presents two key strategies—Non-local Attention Energy Function and Adaptive Update—that collaboratively enhance control precision and realism. On one hand, we theoretically demonstrate that the commonly used attention energy function introduces inherent spatial distribution biases, hindering objects from being uniformly aligned with layout instructions. To overcome this issue, non-local attention prior is explored to redistribute attention scores, facilitating objects to better conform to the specified spatial conditions. On the other hand, we identify that the vanilla backpropagation update rule can cause deviations from the pre-trained domain, leading to out-of-distribution artifacts. We accordingly introduce a Langevin dynamics-based adaptive update scheme as a remedy that promotes in-domain updating while respecting layout constraints. Extensive experiments demonstrate that WinWinLay excels in controlling element placement and achieving photorealistic visual fidelity, outperforming the current state-of-the-art methods.

Blind face restoration (BFR), which involves converting low-quality (LQ) images into high-quality (HQ) images, remains challenging due to complex and unknown degradations. While previous diffusion-based methods utilize feature extractors from LQ images as guidance, using raw LQ images directly as the starting point for the reverse diffusion process offers a theoretically optimal solution. In this work, we propose Pseudo-Hashing Image-to-image Schrödinger Bridge (P-I2SB), a novel framework inspired by optimal mass transport problems, which enhances the restoration potential of Schrödinger Bridge (SB) by correcting data distributions and effectively learning the optimal transport path between any two data distributions. Notably, we theoretically explore and identify that existing methods are limited by the optimality and reversibility of solutions in SB, leading to suboptimal performance. Our approach involves preprocessing HQ images during training by hashing them into pseudo-samples according to a rule related to LQ images, ensuring structural similarity in distribution. This guarantees optimal and reversible solutions in SB, enabling the inference process to learn effectively and allowing P-I2SB to achieve state-of-the-art results in BFR, with more natural textures and retained inference speed compared to previous methods.

Multi-view Diffusion has greatly advanced the development of 3D content creation by generating multiple images from distinct views, achieving remarkable photorealistic results. However, existing works are still vulnerable to inconsistent 3D geometric structures (commonly known as Janus Problem) and severe artifacts. In this paper, we introduce MIRROR, a versatile plug-and-play method that rectifies such inconsistencies in a training-free manner, enabling the acquisition of high-fidelity, realistic structures without compromising diversity. Our key idea focuses on tracing the motion trajectory of physical points across adjacent viewpoints, enabling rectifications based on neighboring observations of the same region. Technically, MIRROR comprises two core modules: Trajectory Tracking Module (TTM) for pixel-wise trajectory tracking that labels identical points across views, and Feature Rectification Module (FRM) for explicitly adjustment of each pixel embedding on noisy synthesized images by minimizing the distance to corresponding block features in neighboring views, thereby achieving consistent outputs. Extensive evaluations demonstrate that MIRROR can seamlessly integrate with a diverse range of off-the-shelf object-level multi-view diffusion models, significantly enhancing both the consistency and the fidelity in an efficient way.

As person image variations are likely to cause a part misalignment problem, most previous person Re-Identification (ReID) works may adopt local feature partition or additional landmark annotations to acquire aligned person features and boost ReID performance. However, such approaches either only achieve coarse-grained part alignments without considering detailed image variations within each part, or require extra annotated landmarks to train an available pose estimation model. In this work, we propose an effective Discriminable and Reliable Transformer (DRFormer) framework to learn part-aligned person representations with only person identity labels. Specifically, the DRFormer framework consists of Discriminable Feature Transformer (DFT) and Reliable Feature Transformer (RFT) modules, which generate discriminable and reliable high-order features, respectively. For reducing the dimension of high-order features, the DFT module utilizes a Self-Attentive Kronecker Product (SAKP) algorithm to promote the representational capabilities of compressed features via a self-attention strategy. For eliminating the background noise, the RFT module mines the foreground regions to adaptively aggregate foreground features via a Gumbel-Softmax strategy. Moreover, the proposed framework derives from an interpretable motivation and elegantly solves part misalignments without using feature partition or pose estimation. This paper theoretically and experimentally demonstrates the superiority of the proposed DRFormer framework, achieving state-of-the-art performance on various person ReID datasets.

Recent works in implicit representations, such as Neural Radiance Fields (NeRF), have advanced the generation of realistic and animatable head avatars from video sequences. These implicit methods are still confronted by visual artifacts and jitters, since the lack of explicit geometric constraints poses a fundamental challenge in accurately modeling complex facial deformations. In this paper, we introduce Dynamic Tetrahedra (DynTet), a novel hybrid representation that encodes explicit dynamic meshes by neural networks to ensure geometric consistency across various motions and viewpoints. DynTet is parameterized by the coordinate-based networks which learn signed distance, deformation, and material texture, anchoring the training data into a predefined tetrahedra grid. Leveraging Marching Tetrahedra, DynTet efficiently decodes textured meshes with a consistent topology, enabling fast rendering through a differentiable rasterizer and supervision via a pixel loss. To enhance training efficiency, we incorporate classical 3D Morphable Models to facilitate geometry learning and define a canonical space for simplifying texture learning. These advantages are readily achievable owing to the effective geometric representation employed in DynTet. Compared with prior works, DynTet demonstrates significant improvements in fidelity, lip synchronization, and real-time performance according to various metrics. Beyond producing stable and visually appealing synthesis videos, our method also outputs the dynamic meshes which is promising to enable many emerging applications. Code is available at https://github.com/zhangzc21/DynTet.

Elastic full-waveform inversion (EFWI) can retrieve multiple subsurface elastic parameters beyond the capabilities of the simple acoustic assumption. Compared to acoustic FWI (AFWI), EFWI faces complexities in dealing with multiple parameters and high nonlinearity in elastic inversion. Wavefield reconstruction inversion (WRI) was proposed to mitigate the cycle skipping and improve the computational efficiency of AFWI. WRI uses the wave equation as a regularization term for the objective of data fitting. By controlling the weight factor for the regularization term, the accuracy of the wave equation is relaxed and the data fitting term is enhanced. Thus, the cycle-skipping issue is reduced. WRI requires wavefield reconstruction in the calculation of the model gradient, which is the key step in WRI. The success of this wavefield reconstruction step highly depends on the accuracy of the source wavelet. In this paper, we propose an elastic WRI (EWRI) with source estimation (SE) method for multiple elastic parameters inversion. In the proposed method, we first reconstruct multicomponent wavefields (vertical and horizontal displacements) and estimate the source wavelet, simultaneously. Then, we formulate the elastic waveform inversion problem into a linear inversion system to mitigate the nonlinearity in EFWI. Applications on synthetic data generated from a modified Overthrust model and a Section of the Sigsbee2A model show the effectiveness of the proposed method in inverting P- and S-wave velocity models with an unknown source wavelet.

In this paper, we focus on analyzing and improving the dropout technique for self-attention layers of Vision Transformer, which is important while surprisingly ignored by prior works. In particular, we conduct researches on three core questions: First, what to drop in self-attention layers? Different from dropping attention weights in literature, we propose to move dropout operations forward ahead of attention matrix calculation and set the Key as the dropout unit, yielding a novel dropout-before-softmax scheme. We theoretically verify that this scheme helps keep both regularization and probability features of attention weights, alleviating the overfittings problem to specific patterns and enhancing the model to globally capture vital information; Second, how to schedule the drop ratio in consecutive layers? In contrast to exploit a constant drop ratio for all layers, we present a new decreasing schedule that gradually decreases the drop ratio along the stack of self-attention layers. We experimentally validate the proposed schedule can avoid overfittings in low-level features and missing in high-level semantics, thus improving the robustness and stableness of model training; Third, whether need to perform structured dropout operation as CNN? We attempt patch-based block-version of dropout operation and find that this useful trick for CNN is not essential for ViT. Given exploration on the above three questions, we present the novel Drop-Key method that regards Key as the drop unit and exploits decreasing schedule for drop ratio, improving ViTs in a general way. Comprehensive experiments demonstrate the effectiveness of DropKey for various ViT architectures, e.g. T2T, VOLO, CeiT and DeiT, as well as for various vision tasks, e.g., image classification, object detection, human-object interaction detection and human body shape recovery.

Blind face restoration (BFR) is important while challenging. Prior works prefer to exploit GAN-based frameworks to tackle this task due to the balance of quality and efficiency. However, these methods suffer from poor stability and adaptability to long-tail distribution, failing to simultaneously retain source identity and restore detail. In this paper, we propose to introduce Diffusion Probabilistic Model (DPM) for BFR to tackle the above problem, given its superiority over GAN in aspects of avoiding training collapse and generating long-tail distribution. We name the proposed framework as DiffBFR. In particular, DiffBFR utilizes a two-step design, that first restores identity information from low-quality images and then enhances texture details according to the distribution of real faces. This design is implemented with two key components: 1) Identity Restoration Module (IRM) for preserving the face details in results. Instead of denoising from pure Gaussian random distribution with LQ images as the condition during the reverse process, we propose a novel truncated sampling method which starts from LQ images with part noise added. We theoretically prove that this change shrinks the evidence lower bound of DPM and then restores more original details. With theoretical proof, two cascade conditional DPMs with different input sizes are introduced to strengthen this sampling effect and reduce training difficulty in the high-resolution image generated directly. 2) Texture Enhancement Module (TEM) for polishing the texture of the image. Here an unconditional DPM, a LQ-free model, is introduced to further force the restorations to appear realistic. We theoretically proved that this unconditional DPM trained on pure HQ images contributes to justifying the correct distribution of inference images output from IRM in pixel-level space. Concretely, truncated sampling with fractional time step is utilized to polish pixel-level textures while preserving identity information. Our experiments demonstrated that the proposed DiffBFR achieves significantly superior results to state-of-the-art methods both quantitatively and qualitatively.

Existing works have advanced Text-to-Image (TTI) diffusion models for video editing in a one-shot learning manner. Despite their low requirements of data and computation, these methods might produce results of unsatisfied consistency with text prompt as well as temporal sequence, limiting their applications in the real world. In this paper, we propose to address the above issues with a novel EI$^2$ model towards Enhancing vIdeo Editing consIstency of TTI-based frameworks. Specifically, we analyze and find that the inconsistent problem is caused by newly added modules into TTI models for learning temporal information. These modules lead to covariate shift in the feature space, which harms the editing capability. Thus, we design EI$^2$ to tackle the above drawbacks with two classical modules: Shift-restricted Temporal Attention Module (STAM) and Fine-coarse Frame Attention Module (FFAM). First, through theoretical analysis, we demonstrate that covariate shift is highly related to Layer Normalization, thus STAM employs a Instance Centering layer replacing it to preserve the distribution of temporal features. In addition, STAM employs an attention layer with normalized mapping to transform temporal features while constraining the variance shift. As the second part, we incorporate STAM with a novel FFAM, which efficiently leverages fine-coarse spatial information of overall frames to further enhance temporal consistency. Extensive experiments demonstrate the superiority of the proposed EI$^2$ model.

Spatiotemporal modeling in an unified architecture is key for video action recognition. This paper proposes a Shrinking Temporal Attention Transformer (STAT), which efficiently builts spatiotemporal attention maps considering the attenuation of spatial attention in short and long temporal sequences. Specifically, for short-term temporal tokens, query token interacts with them in a fine-grained manner in dealing with short-range motion. It then shrinks to a coarse attention in neighborhood for long-term tokens, to provide larger receptive field for long-range spatial aggregation. Both of them are composed in a short-long temporal integrated block to build visual appearances and temporal structure concurrently with lower costly in computation. We conduct thorough ablation studies, and achieve state-of-the-art results on multiple action recognition benchmarks including Kinetics400 and Something-Something v2, outperforming prior methods with 50% less FLOPs and without any pretrained model.

Generative based strategy has shown great potential in the Generalized Zero-Shot Learning task. However, it suffers severe generalization problem due to lacking of feature diversity for unseen classes to train a good classifier. In this paper, we propose to enhance the generalizability of GZSL models via improving feature diversity of unseen classes. For this purpose, we present a novel Diverse Feature Synthesis (DFS) model. Different from prior works that solely utilize semantic knowledge in the generation process, DFS leverages visual knowledge with semantic one in a unified way, thus deriving class-specific diverse feature samples and leading to robust classifier for recognizing both seen and unseen classes in the testing phase. To simplify the learning, DFS represents visual and semantic knowledge in the aligned space, making it able to produce good feature samples with a low-complexity implementation. Accordingly, DFS is composed of two consecutive generators: an aligned feature generator, transferring semantic and visual representations into aligned features; a synthesized feature generator, producing diverse feature samples of unseen classes in the aligned space. We conduct comprehensive experiments to verify the efficacy of DFS. Results demonstrate its effectiveness to generate diverse features for unseen classes, leading to superior performance on multiple benchmarks. Code will be released upon acceptance.

This paper presents two linked empirical studies focused on uncertainty visualization. The experiments are framed from two conceptual perspectives. First, a typology of uncertainty is used to delineate kinds of uncertainty matched with space, time, and attribute components of data. Second, concepts from visual semiotics are applied to characterize the kind of visual signification that is appropriate for representing those different categories of uncertainty. This framework guided the two experiments reported here. The first addresses representation intuitiveness, considering both visual variables and iconicity of representation. The second addresses relative performance of the most intuitive abstract and iconic representations of uncertainty on a map reading task. Combined results suggest initial guidelines for representing uncertainty and discussion focuses on practical applicability of results.