Blind super-resolution methods based on stable diffusion showcase formidable generative capabilities in reconstructing clear high-resolution images with intricate details from low-resolution inputs. However, their practical applicability is often hampered by poor efficiency, stemming from the requirement of thousands or hundreds of sampling steps. Inspired by the efficient adversarial diffusion distillation (ADD), we design~\name~to address this issue by incorporating the ideas of both distillation and ControlNet. Specifically, we first propose a prediction-based self-refinement strategy to provide high-frequency information in the student model output with marginal additional time cost. Furthermore, we refine the training process by employing HR images, rather than LR images, to regulate the teacher model, providing a more robust constraint for distillation. Second, we introduce a timestep-adaptive ADD to address the perception-distortion imbalance problem introduced by original ADD. Extensive experiments demonstrate our~\name~generates better restoration results, while achieving faster speed than previous SD-based state-of-the-art models (e.g., $7$$\times$ faster than SeeSR).
相關內容
In causal inference, estimating Heterogeneous Treatment Effects (HTEs) from observational data is critical for understanding how different subgroups respond to treatments, with broad applications such as precision medicine and targeted advertising. However, existing work on HTE, subgroup discovery, and causal visualization is insufficient to address two challenges: first, the sheer number of potential subgroups and the necessity to balance multiple objectives (e.g., high effects and low variances) pose a considerable analytical challenge. Second, effective subgroup analysis has to follow the analysis goal specified by users and provide causal results with verification. To this end, we propose a visual analytics approach for subgroup-based causal heterogeneity exploration. Specifically, we first formulate causal subgroup discovery as a constrained multi-objective optimization problem and adopt a heuristic genetic algorithm to learn the Pareto front of optimal subgroups described by interpretable rules. Combining with this model, we develop a prototype system, CausalPrism, that incorporates tabular visualization, multi-attribute rankings, and uncertainty plots to support users in interactively exploring and sorting subgroups and explaining treatment effects. Quantitative experiments validate that the proposed model can efficiently mine causal subgroups that outperform state-of-the-art HTE and subgroup discovery methods, and case studies and expert interviews demonstrate the effectiveness and usability of the system. Code is available at //osf.io/jaqmf/?view_only=ac9575209945476b955bf829c85196e9.
Traditional image processing methods employing partial differential equations (PDEs) offer a multitude of meaningful regularizers, along with valuable theoretical foundations for a wide range of image-related tasks. This makes their integration into neural networks a promising avenue. In this paper, we introduce a novel regularization approach inspired by the reverse process of PDE-based evolution models. Specifically, we propose inverse evolution layers (IELs), which serve as bad property amplifiers to penalize neural networks of which outputs have undesired characteristics. Using IELs, one can achieve specific regularization objectives and endow neural networks' outputs with corresponding properties of the PDE models. Our experiments, focusing on semantic segmentation tasks using heat-diffusion IELs, demonstrate their effectiveness in mitigating noisy label effects. Additionally, we develop curve-motion IELs to enforce convex shape regularization in neural network-based segmentation models for preventing the generation of concave outputs. Theoretical analysis confirms the efficacy of IELs as an effective regularization mechanism, particularly in handling training with label issues.
4D head capture aims to generate dynamic topological meshes and corresponding texture maps from videos, which is widely utilized in movies and games for its ability to simulate facial muscle movements and recover dynamic textures in pore-squeezing. The industry often adopts the method involving multi-view stereo and non-rigid alignment. However, this approach is prone to errors and heavily reliant on time-consuming manual processing by artists. To simplify this process, we propose Topo4D, a novel framework for automatic geometry and texture generation, which optimizes densely aligned 4D heads and 8K texture maps directly from calibrated multi-view time-series images. Specifically, we first represent the time-series faces as a set of dynamic 3D Gaussians with fixed topology in which the Gaussian centers are bound to the mesh vertices. Afterward, we perform alternative geometry and texture optimization frame-by-frame for high-quality geometry and texture learning while maintaining temporal topology stability. Finally, we can extract dynamic facial meshes in regular wiring arrangement and high-fidelity textures with pore-level details from the learned Gaussians. Extensive experiments show that our method achieves superior results than the current SOTA face reconstruction methods both in the quality of meshes and textures. Project page: //xuanchenli.github.io/Topo4D/.
Large language models (LLMs) exhibit impressive capabilities across a wide range of tasks, yet the choice of which model to use often involves a trade-off between performance and cost. More powerful models, though effective, come with higher expenses, while less capable models are more cost-effective. To address this dilemma, we propose several efficient router models that dynamically select between a stronger and a weaker LLM during inference, aiming to optimize the balance between cost and response quality. We develop a training framework for these routers leveraging human preference data and data augmentation techniques to enhance performance. Our evaluation on widely-recognized benchmarks shows that our approach significantly reduces costs-by over 2 times in certain cases-without compromising the quality of responses. Interestingly, our router models also demonstrate significant transfer learning capabilities, maintaining their performance even when the strong and weak models are changed at test time. This highlights the potential of these routers to provide a cost-effective yet high-performance solution for deploying LLMs.
Autonomous exploration requires a robot to explore an unknown environment while constructing an accurate map using Simultaneous Localization and Mapping (SLAM) techniques. Without prior information, the exploration performance is usually conservative due to the limited planning horizon. This paper exploits prior information about the environment, represented as a topo-metric graph, to benefit both the exploration efficiency and the pose graph reliability in SLAM. Based on the relationship between pose graph reliability and graph topology, we formulate a SLAM-aware path planning problem over the prior graph, which finds a fast exploration path enhanced with the globally informative loop-closing actions to stabilize the SLAM pose graph. A greedy algorithm is proposed to solve the problem, where theoretical thresholds are derived to significantly prune non-optimal loop-closing actions, without affecting the potential informative ones. Furthermore, we incorporate the proposed planner into a hierarchical exploration framework, with flexible features including path replanning, and online prior graph update that adds additional information to the prior graph. Simulation and real-world experiments indicate that the proposed method can reliably achieve higher mapping accuracy than compared methods when exploring environments with rich topologies, while maintaining comparable exploration efficiency. Our method has been open-sourced on GitHub.
For robust visual-inertial SLAM in perceptually-challenging indoor environments,recent studies exploit line features to extract descriptive information about scene structure to deal with the degeneracy of point features. But existing point-line-based SLAM methods mainly use Pl\"ucker matrix or orthogonal representation to represent a line, which needs to calculate at least four variables to determine a line. Given the numerous line features to determine in each frame, the overly flexible line representation increases the computation burden and comprises the accuracy of the results. In this paper, we propose inverse depth representation for a line, which models each extracted line feature using only two variables, i.e., the inverse depths of the two ending points. It exploits the fact that the projected line's pixel coordinates on the image plane are rather accurate, which partially restrict the line. Using this compact line presentation, Inverse Depth Line SLAM (IDLS) is proposed to track the line features in SLAM in an accurate and efficient way. A robust line triangulation method and a novel line re-projection error model are introduced. And a two-step optimization method is proposed to firstly determine the lines and then to estimate the camera poses in each frame. IDLS is extensively evaluated in multiple perceptually-challenging datasets. The results show it is more accurate, robust, and needs lower computational overhead than the current state-of-the-art of point-line-based SLAM methods.
Motion-based controllable text-to-video generation involves motions to control the video generation. Previous methods typically require the training of models to encode motion cues or the fine-tuning of video diffusion models. However, these approaches often result in suboptimal motion generation when applied outside the trained domain. In this work, we propose MotionClone, a training-free framework that enables motion cloning from a reference video to control text-to-video generation. We employ temporal attention in video inversion to represent the motions in the reference video and introduce primary temporal-attention guidance to mitigate the influence of noisy or very subtle motions within the attention weights. Furthermore, to assist the generation model in synthesizing reasonable spatial relationships and enhance its prompt-following capability, we propose a location-aware semantic guidance mechanism that leverages the coarse location of the foreground from the reference video and original classifier-free guidance features to guide the video generation. Extensive experiments demonstrate that MotionClone exhibits proficiency in both global camera motion and local object motion, with notable superiority in terms of motion fidelity, textual alignment, and temporal consistency.
The unique complementarity of frame-based and event cameras for high frame rate object tracking has recently inspired some research attempts to develop multi-modal fusion approaches. However, these methods directly fuse both modalities and thus ignore the environmental attributes, e.g., motion blur, illumination variance, occlusion, scale variation, etc. Meanwhile, no interaction between search and template features makes distinguishing target objects and backgrounds difficult. As a result, performance degradation is induced especially in challenging conditions. This paper proposes a novel and effective Transformer-based event-guided tracking framework, called eMoE-Tracker, which achieves new SOTA performance under various conditions. Our key idea is to disentangle the environment into several learnable attributes to dynamically learn the attribute-specific features for better interaction and discriminability between the target information and background. To achieve the goal, we first propose an environmental Mix-of-Experts (eMoE) module that is built upon the environmental Attributes Disentanglement to learn attribute-specific features and environmental Attributes Gating to assemble the attribute-specific features by the learnable attribute scores dynamically. The eMoE module is a subtle router that fine-tunes the transformer backbone more efficiently. We then introduce a contrastive relation modeling (CRM) module to improve interaction and discriminability between the target information and background. Extensive experiments on diverse event-based benchmark datasets showcase the superior performance of our eMoE-Tracker compared to the prior arts.
Video instance segmentation requires detecting, segmenting, and tracking objects in videos, typically relying on costly video annotations. This paper introduces a method that eliminates video annotations by utilizing image datasets. The PM-VIS algorithm is adapted to handle both bounding box and instance-level pixel annotations dynamically. We introduce ImageNet-bbox to supplement missing categories in video datasets and propose the PM-VIS+ algorithm to adjust supervision based on annotation types. To enhance accuracy, we use pseudo masks and semi-supervised optimization techniques on unannotated video data. This method achieves high video instance segmentation performance without manual video annotations, offering a cost-effective solution and new perspectives for video instance segmentation applications. The code will be available in //github.com/ldknight/PM-VIS-plus
InteractE: Improving Convolution-based Knowledge Graph Embeddings by Increasing Feature Interactions
Most existing knowledge graphs suffer from incompleteness, which can be alleviated by inferring missing links based on known facts. One popular way to accomplish this is to generate low-dimensional embeddings of entities and relations, and use these to make inferences. ConvE, a recently proposed approach, applies convolutional filters on 2D reshapings of entity and relation embeddings in order to capture rich interactions between their components. However, the number of interactions that ConvE can capture is limited. In this paper, we analyze how increasing the number of these interactions affects link prediction performance, and utilize our observations to propose InteractE. InteractE is based on three key ideas -- feature permutation, a novel feature reshaping, and circular convolution. Through extensive experiments, we find that InteractE outperforms state-of-the-art convolutional link prediction baselines on FB15k-237. Further, InteractE achieves an MRR score that is 9%, 7.5%, and 23% better than ConvE on the FB15k-237, WN18RR and YAGO3-10 datasets respectively. The results validate our central hypothesis -- that increasing feature interaction is beneficial to link prediction performance. We make the source code of InteractE available to encourage reproducible research.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191