Dong Xu

Zicheng Wang, Zhen Zhao, Xiaoxia Xing et al.

Semi-supervised semantic segmentation (SSS) has recently gained increasing research interest as it can reduce the requirement for large-scale fully-annotated training data. The current methods often suffer from the confirmation bias from the pseudo-labelling process, which can be alleviated by the co-training framework. The current co-training-based SSS methods rely on hand-crafted perturbations to prevent the different sub-nets from collapsing into each other, but these artificial perturbations cannot lead to the optimal solution. In this work, we propose a new conflict-based cross-view consistency (CCVC) method based on a two-branch co-training framework which aims at enforcing the two sub-nets to learn informative features from irrelevant views. In particular, we first propose a new cross-view consistency (CVC) strategy that encourages the two sub-nets to learn distinct features from the same input by introducing a feature discrepancy loss, while these distinct features are expected to generate consistent prediction scores of the input. The CVC strategy helps to prevent the two sub-nets from stepping into the collapse. In addition, we further propose a conflict-based pseudo-labelling (CPL) method to guarantee the model will learn more useful information from conflicting predictions, which will lead to a stable training process. We validate our new CCVC approach on the SSS benchmark datasets where our method achieves new state-of-the-art performance. Our code is available at https://github.com/xiaoyao3302/CCVC.

Lichen Zhao, Daigang Cai, Jing Zhang et al.

Recently, 3D vision-and-language tasks have attracted increasing research interest. Compared to other vision-and-language tasks, the 3D visual question answering (VQA) task is less exploited and is more susceptible to language priors and co-reference ambiguity. Meanwhile, a couple of recently proposed 3D VQA datasets do not well support 3D VQA task due to their limited scale and annotation methods. In this work, we formally define and address a 3D grounded VQA task by collecting a new 3D VQA dataset, referred to as FE-3DGQA, with diverse and relatively free-form question-answer pairs, as well as dense and completely grounded bounding box annotations. To achieve more explainable answers, we labelled the objects appeared in the complex QA pairs with different semantic types, including answer-grounded objects (both appeared and not appeared in the questions), and contextual objects for answer-grounded objects. We also propose a new 3D VQA framework to effectively predict the completely visually grounded and explainable answer. Extensive experiments verify that our newly collected benchmark datasets can be effectively used to evaluate various 3D VQA methods from different aspects and our newly proposed framework also achieves state-of-the-art performance on the new benchmark dataset. Both the newly collected dataset and our codes will be publicly available at http://github.com/zlccccc/3DGQA.

Ziqin Wang, Bowen Cheng, Lichen Zhao et al.

The task of 3D semantic scene graph (3DSSG) prediction in the point cloud is challenging since (1) the 3D point cloud only captures geometric structures with limited semantics compared to 2D images, and (2) long-tailed relation distribution inherently hinders the learning of unbiased prediction. Since 2D images provide rich semantics and scene graphs are in nature coped with languages, in this study, we propose Visual-Linguistic Semantics Assisted Training (VL-SAT) scheme that can significantly empower 3DSSG prediction models with discrimination about long-tailed and ambiguous semantic relations. The key idea is to train a powerful multi-modal oracle model to assist the 3D model. This oracle learns reliable structural representations based on semantics from vision, language, and 3D geometry, and its benefits can be heterogeneously passed to the 3D model during the training stage. By effectively utilizing visual-linguistic semantics in training, our VL-SAT can significantly boost common 3DSSG prediction models, such as SGFN and SGGpoint, only with 3D inputs in the inference stage, especially when dealing with tail relation triplets. Comprehensive evaluations and ablation studies on the 3DSSG dataset have validated the effectiveness of the proposed scheme. Code is available at https://github.com/wz7in/CVPR2023-VLSAT.

Ziming Wang, Xiaoliang Huo, Zhenghao Chen et al.

Point cloud registration aims at estimating the geometric transformation between two point cloud scans, in which point-wise correspondence estimation is the key to its success. In addition to previous methods that seek correspondences by hand-crafted or learnt geometric features, recent point cloud registration methods have tried to apply RGB-D data to achieve more accurate correspondence. However, it is not trivial to effectively fuse the geometric and visual information from these two distinctive modalities, especially for the registration problem. In this work, we propose a new Geometry-Aware Visual Feature Extractor (GAVE) that employs multi-scale local linear transformation to progressively fuse these two modalities, where the geometric features from the depth data act as the geometry-dependent convolution kernels to transform the visual features from the RGB data. The resultant visual-geometric features are in canonical feature spaces with alleviated visual dissimilarity caused by geometric changes, by which more reliable correspondence can be achieved. The proposed GAVE module can be readily plugged into recent RGB-D point cloud registration framework. Extensive experiments on 3D Match and ScanNet demonstrate that our method outperforms the state-of-the-art point cloud registration methods even without correspondence or pose supervision. The code is available at: https://github.com/514DNA/LLT.

Weiqi Sun, Rui Su, Qian Yu et al. · amazon-science, princeton

Weakly supervised temporal action localization (WTAL) aims to localize actions in untrimmed videos with only weak supervision information (e.g. video-level labels). Most existing models handle all input videos with a fixed temporal scale. However, such models are not sensitive to actions whose pace of the movements is different from the ``normal" speed, especially slow-motion action instances, which complete the movements with a much slower speed than their counterparts with a normal speed. Here arises the slow-motion blurred issue: It is hard to explore salient slow-motion information from videos at ``normal" speed. In this paper, we propose a novel framework termed Slow Motion Enhanced Network (SMEN) to improve the ability of a WTAL network by compensating its sensitivity on slow-motion action segments. The proposed SMEN comprises a Mining module and a Localization module. The mining module generates mask to mine slow-motion-related features by utilizing the relationships between the normal motion and slow motion; while the localization module leverages the mined slow-motion features as complementary information to improve the temporal action localization results. Our proposed framework can be easily adapted by existing WTAL networks and enable them be more sensitive to slow-motion actions. Extensive experiments on three benchmarks are conducted, which demonstrate the high performance of our proposed framework.

Feiyu Wang, Qin Wang, Wen Li et al. · eth-zurich

In this work, we revisit the semi-supervised learning (SSL) problem from a new perspective of explicitly reducing empirical distribution mismatch between labeled and unlabeled samples. Benefited from this new perspective, we first propose a new deep semi-supervised learning framework called Semi-supervised Learning by Empirical Distribution Alignment (SLEDA), in which existing technologies from the domain adaptation community can be readily used to address the semi-supervised learning problem through reducing the empirical distribution distance between labeled and unlabeled data. Based on this framework, we also develop a new theoretical generalization bound for the research community to better understand the semi-supervised learning problem, in which we show the generalization error of semi-supervised learning can be effectively bounded by minimizing the training error on labeled data and the empirical distribution distance between labeled and unlabeled data. Building upon our new framework and the theoretical bound, we develop a simple and effective deep semi-supervised learning method called Augmented Distribution Alignment Network (ADA-Net) by simultaneously adopting the well-established adversarial training strategy from the domain adaptation community and a simple sample interpolation strategy for data augmentation. Additionally, we incorporate both strategies in our ADA-Net into two exiting SSL methods to further improve their generalization capability, which indicates that our new framework provides a complementary solution for solving the SSL problem. Our comprehensive experimental results on two benchmark datasets SVHN and CIFAR-10 for the semi-supervised image recognition task and another two benchmark datasets ModelNet40 and ShapeNet55 for the semi-supervised point cloud recognition task demonstrate the effectiveness of our proposed framework for SSL.

Jianing Qiu, Lin Li, Jiankai Sun et al.

Large AI models, or foundation models, are models recently emerging with massive scales both parameter-wise and data-wise, the magnitudes of which can reach beyond billions. Once pretrained, large AI models demonstrate impressive performance in various downstream tasks. A prime example is ChatGPT, whose capability has compelled people's imagination about the far-reaching influence that large AI models can have and their potential to transform different domains of our lives. In health informatics, the advent of large AI models has brought new paradigms for the design of methodologies. The scale of multi-modal data in the biomedical and health domain has been ever-expanding especially since the community embraced the era of deep learning, which provides the ground to develop, validate, and advance large AI models for breakthroughs in health-related areas. This article presents a comprehensive review of large AI models, from background to their applications. We identify seven key sectors in which large AI models are applicable and might have substantial influence, including 1) bioinformatics; 2) medical diagnosis; 3) medical imaging; 4) medical informatics; 5) medical education; 6) public health; and 7) medical robotics. We examine their challenges, followed by a critical discussion about potential future directions and pitfalls of large AI models in transforming the field of health informatics.

Erjian Guo, Huazhu Fu, Luping Zhou et al.

Deep learning based image enhancement models have largely improved the readability of fundus images in order to decrease the uncertainty of clinical observations and the risk of misdiagnosis. However, due to the difficulty of acquiring paired real fundus images at different qualities, most existing methods have to adopt synthetic image pairs as training data. The domain shift between the synthetic and the real images inevitably hinders the generalization of such models on clinical data. In this work, we propose an end-to-end optimized teacher-student framework to simultaneously conduct image enhancement and domain adaptation. The student network uses synthetic pairs for supervised enhancement, and regularizes the enhancement model to reduce domain-shift by enforcing teacher-student prediction consistency on the real fundus images without relying on enhanced ground-truth. Moreover, we also propose a novel multi-stage multi-attention guided enhancement network (MAGE-Net) as the backbones of our teacher and student network. Our MAGE-Net utilizes multi-stage enhancement module and retinal structure preservation module to progressively integrate the multi-scale features and simultaneously preserve the retinal structures for better fundus image quality enhancement. Comprehensive experiments on both real and synthetic datasets demonstrate that our framework outperforms the baseline approaches. Moreover, our method also benefits the downstream clinical tasks.

Zhihao Hu, Guo Lu, Jinyang Guo et al.

The previous deep video compression approaches only use the single scale motion compensation strategy and rarely adopt the mode prediction technique from the traditional standards like H.264/H.265 for both motion and residual compression. In this work, we first propose a coarse-to-fine (C2F) deep video compression framework for better motion compensation, in which we perform motion estimation, compression and compensation twice in a coarse to fine manner. Our C2F framework can achieve better motion compensation results without significantly increasing bit costs. Observing hyperprior information (i.e., the mean and variance values) from the hyperprior networks contains discriminant statistical information of different patches, we also propose two efficient hyperprior-guided mode prediction methods. Specifically, using hyperprior information as the input, we propose two mode prediction networks to respectively predict the optimal block resolutions for better motion coding and decide whether to skip residual information from each block for better residual coding without introducing additional bit cost while bringing negligible extra computation cost. Comprehensive experimental results demonstrate our proposed C2F video compression framework equipped with the new hyperprior-guided mode prediction methods achieves the state-of-the-art performance on HEVC, UVG and MCL-JCV datasets.

Zhiye Guo, Jian Liu, Yanli Wang et al.

Denoising diffusion models have emerged as one of the most powerful generative models in recent years. They have achieved remarkable success in many fields, such as computer vision, natural language processing (NLP), and bioinformatics. Although there are a few excellent reviews on diffusion models and their applications in computer vision and NLP, there is a lack of an overview of their applications in bioinformatics. This review aims to provide a rather thorough overview of the applications of diffusion models in bioinformatics to aid their further development in bioinformatics and computational biology. We start with an introduction of the key concepts and theoretical foundations of three cornerstone diffusion modeling frameworks (denoising diffusion probabilistic models, noise-conditioned scoring networks, and stochastic differential equations), followed by a comprehensive description of diffusion models employed in the different domains of bioinformatics, including cryo-EM data enhancement, single-cell data analysis, protein design and generation, drug and small molecule design, and protein-ligand interaction. The review is concluded with a summary of the potential new development and applications of diffusion models in bioinformatics.

Jinglong Wang, Xiawei Li, Jing Zhang et al.

The pre-trained text-image discriminative models, such as CLIP, has been explored for open-vocabulary semantic segmentation with unsatisfactory results due to the loss of crucial localization information and awareness of object shapes. Recently, there has been a growing interest in expanding the application of generative models from generation tasks to semantic segmentation. These approaches utilize generative models either for generating annotated data or extracting features to facilitate semantic segmentation. This typically involves generating a considerable amount of synthetic data or requiring additional mask annotations. To this end, we uncover the potential of generative text-to-image diffusion models (e.g., Stable Diffusion) as highly efficient open-vocabulary semantic segmenters, and introduce a novel training-free approach named DiffSegmenter. The insight is that to generate realistic objects that are semantically faithful to the input text, both the complete object shapes and the corresponding semantics are implicitly learned by diffusion models. We discover that the object shapes are characterized by the self-attention maps while the semantics are indicated through the cross-attention maps produced by the denoising U-Net, forming the basis of our segmentation results.Additionally, we carefully design effective textual prompts and a category filtering mechanism to further enhance the segmentation results. Extensive experiments on three benchmark datasets show that the proposed DiffSegmenter achieves impressive results for open-vocabulary semantic segmentation.

Ximing Xing, Chuang Wang, Haitao Zhou et al.

Even though trained mainly on images, we discover that pretrained diffusion models show impressive power in guiding sketch synthesis. In this paper, we present DiffSketcher, an innovative algorithm that creates \textit{vectorized} free-hand sketches using natural language input. DiffSketcher is developed based on a pre-trained text-to-image diffusion model. It performs the task by directly optimizing a set of Bézier curves with an extended version of the score distillation sampling (SDS) loss, which allows us to use a raster-level diffusion model as a prior for optimizing a parametric vectorized sketch generator. Furthermore, we explore attention maps embedded in the diffusion model for effective stroke initialization to speed up the generation process. The generated sketches demonstrate multiple levels of abstraction while maintaining recognizability, underlying structure, and essential visual details of the subject drawn. Our experiments show that DiffSketcher achieves greater quality than prior work. The code and demo of DiffSketcher can be found at https://ximinng.github.io/DiffSketcher-project/.

Chenjian Gao, Qian Yu, Lu Sheng et al.

Reconstructing a 3D shape based on a single sketch image is challenging due to the large domain gap between a sparse, irregular sketch and a regular, dense 3D shape. Existing works try to employ the global feature extracted from sketch to directly predict the 3D coordinates, but they usually suffer from losing fine details that are not faithful to the input sketch. Through analyzing the 3D-to-2D projection process, we notice that the density map that characterizes the distribution of 2D point clouds (i.e., the probability of points projected at each location of the projection plane) can be used as a proxy to facilitate the reconstruction process. To this end, we first translate a sketch via an image translation network to a more informative 2D representation that can be used to generate a density map. Next, a 3D point cloud is reconstructed via a two-stage probabilistic sampling process: first recovering the 2D points (i.e., the x and y coordinates) by sampling the density map; and then predicting the depth (i.e., the z coordinate) by sampling the depth values at the ray determined by each 2D point. Extensive experiments are conducted, and both quantitative and qualitative results show that our proposed approach significantly outperforms other baseline methods.

Dong Xu, Jialun Cao, Guozhao Mo et al.

Formal specification is essential for rigorous program verification, yet writing correct specifications remains costly and difficult to automate. Although large language models (LLMs) and agents have shown promising progress, their true capabilities and failure modes remain unclear. We present the first systematic and contamination-aware study of LLM- and agent-based formal specification generation for C programs. We introduce LiveFMBench, a continuously evolving benchmark of 630 ACSL (ANSI/ISO C Specification Language)-annotated C programs, including 360 newly collected cases designed to mitigate data leakage. Using this benchmark, we evaluate direct prompting with different sampling sizes, reasoning-enabled (thinking mode) inference, the agentic pipeline, and perform a fine-grained failure analysis. Experimental results reveal that naive evaluation substantially overestimates performance because models under direct prompting may exhibit unfaithful behaviors, such as deceiving automated provers or ignoring code-context constraints; after excluding such cases, the true specification generation accuracy drops by approximately 20\%. We further find that both increased sampling and thinking mode significantly improve success rates, with smaller models benefiting more from thinking mode. Agentic pipelines are particularly effective under low sampling budgets and on harder datasets. Failure analysis further shows that incorrect loop invariants are the dominant error type, while agentic pipelines notably reduce assertion errors. These results expose fundamental limitations in current LLM-based approaches and suggest they remain far from replacing human-authored formal specifications. We release LiveFMBench at https://huggingface.co/datasets/fm-universe/Live-FM-Bench and all evaluation artifacts to support future research.

Maomao Li, Ge Yuan, Cairong Wang et al.

This paper proposes a novel approach to face swapping from the perspective of fine-grained facial editing, dubbed "editing for swapping" (E4S). The traditional face swapping methods rely on global feature extraction and fail to preserve the detailed source identity. In contrast, we propose a Regional GAN Inversion (RGI) method, which allows the explicit disentanglement of shape and texture. Specifically, our E4S performs face swapping in the latent space of a pretrained StyleGAN, where a multi-scale mask-guided encoder is applied to project the texture of each facial component into regional style codes and a mask-guided injection module manipulating feature maps with the style codes. Based on this disentanglement, face swapping can be simplified as style and mask swapping. Besides, due to the large lighting condition gap, transferring the source skin into the target image may lead to disharmony lighting. We propose a re-coloring network to make the swapped face maintain the target lighting condition while preserving the source skin. Further, to deal with the potential mismatch areas during mask exchange, we design a face inpainting module to refine the face shape. The extensive comparisons with state-of-the-art methods demonstrate that our E4S outperforms existing methods in preserving texture, shape, and lighting. Our implementation is available at https://github.com/e4s2024/E4S2024.

Yubin Xiao, Di Wang, Boyang Li et al.

The Traveling Salesman Problem (TSP) is a well-known combinatorial optimization problem with broad real-world applications. Recently, neural networks have gained popularity in this research area because as shown in the literature, they provide strong heuristic solutions to TSPs. Compared to autoregressive neural approaches, non-autoregressive (NAR) networks exploit the inference parallelism to elevate inference speed but suffer from comparatively low solution quality. In this paper, we propose a novel NAR model named NAR4TSP, which incorporates a specially designed architecture and an enhanced reinforcement learning strategy. To the best of our knowledge, NAR4TSP is the first TSP solver that successfully combines RL and NAR networks. The key lies in the incorporation of NAR network output decoding into the training process. NAR4TSP efficiently represents TSP encoded information as rewards and seamlessly integrates it into reinforcement learning strategies, while maintaining consistent TSP sequence constraints during both training and testing phases. Experimental results on both synthetic and real-world TSPs demonstrate that NAR4TSP outperforms five state-of-the-art models in terms of solution quality, inference speed, and generalization to unseen scenarios.

Yinjie Zhao, Lichen Zhao, Qian Yu et al.

With the emergence of VR and AR, 360° data attracts increasing attention from the computer vision and multimedia communities. Typically, 360° data is projected into 2D ERP (equirectangular projection) images for feature extraction. However, existing methods cannot handle the distortions that result from the projection, hindering the development of 360-data-based tasks. Therefore, in this paper, we propose a Transformer-based model called DATFormer to address the distortion problem. We tackle this issue from two perspectives. Firstly, we introduce two distortion-adaptive modules. The first is a Distortion Mapping Module, which guides the model to pre-adapt to distorted features globally. The second module is a Distortion-Adaptive Attention Block that reduces local distortions on multi-scale features. Secondly, to exploit the unique characteristics of 360° data, we present a learnable relation matrix and use it as part of the positional embedding to further improve performance. Extensive experiments are conducted on three public datasets, and the results show that our model outperforms existing 2D SOD (salient object detection) and 360 SOD methods.

Zhihao Hu, Dong Xu

Recently, diffusion models like StableDiffusion have achieved impressive image generation results. However, the generation process of such diffusion models is uncontrollable, which makes it hard to generate videos with continuous and consistent content. In this work, by using the diffusion model with ControlNet, we proposed a new motion-guided video-to-video translation framework called VideoControlNet to generate various videos based on the given prompts and the condition from the input video. Inspired by the video codecs that use motion information for reducing temporal redundancy, our framework uses motion information to prevent the regeneration of the redundant areas for content consistency. Specifically, we generate the first frame (i.e., the I-frame) by using the diffusion model with ControlNet. Then we generate other key frames (i.e., the P-frame) based on the previous I/P-frame by using our newly proposed motion-guided P-frame generation (MgPG) method, in which the P-frames are generated based on the motion information and the occlusion areas are inpainted by using the diffusion model. Finally, the rest frames (i.e., the B-frame) are generated by using our motion-guided B-frame interpolation (MgBI) module. Our experiments demonstrate that our proposed VideoControlNet inherits the generation capability of the pre-trained large diffusion model and extends the image diffusion model to the video diffusion model by using motion information. More results are provided at our project page.

Ximing Xing, Haitao Zhou, Chuang Wang et al.

Text-guided scalable vector graphics (SVG) synthesis has broad applications in icon and sketch generation. However, existing text-to-SVG methods often suffer from limited editability, suboptimal visual quality, and low sample diversity. To address these challenges, we propose \textbf{SVGDreamer}, a novel framework for text-guided vector graphics synthesis. Our method introduces a \textbf{semantic-driven image vectorization (SIVE)} process, which decomposes the generation procedure into foreground objects and background elements, thereby improving structural controllability and editability. In particular, SIVE incorporates attention-based primitive control and an attention-mask loss to facilitate fine-grained manipulation of individual vector elements. To further improve generation quality and diversity, we propose \textbf{Vectorized Particle-based Score Distillation (VPSD)}, which models SVGs as distributions over control points and colors. Compared with existing text-to-SVG optimization methods, VPSD alleviates over-smoothed shapes, over-saturated colors, limited diversity, and slow convergence. Moreover, VPSD leverages a reward model to reweight vector particles, leading to better visual aesthetics and faster convergence. Extensive experiments demonstrate that SVGDreamer consistently outperforms existing baselines in editability, visual quality, and diversity. Project page: https://ximinng.github.io/SVGDreamer-project/

MiniCPM Team, Wenhao An, Yingfa Chen et al. · tsinghua

The evolution of large language models (LLMs) towards applications with ultra-long contexts faces challenges posed by the high computational and memory costs of the Transformer architecture. While existing sparse and linear attention mechanisms attempt to mitigate these issues, they typically involve a trade-off between memory efficiency and model performance. This paper introduces MiniCPM-SALA, a 9B-parameter hybrid architecture that integrates the high-fidelity long-context modeling of sparse attention (InfLLM-V2) with the global efficiency of linear attention (Lightning Attention). By employing a layer selection algorithm to integrate these mechanisms in a 1:3 ratio and utilizing a hybrid positional encoding (HyPE), the model maintains efficiency and performance for long-context tasks. Furthermore, we introduce a cost-effective continual training framework that transforms pre-trained Transformer-based models into hybrid models, which reduces training costs by approximately 75% compared to training from scratch. Extensive experiments show that MiniCPM-SALA maintains general capabilities comparable to full-attention models while offering improved efficiency. On a single NVIDIA A6000D GPU, the model achieves up to 3.5x the inference speed of the full-attention model at the sequence length of 256K tokens and supports context lengths of up to 1M tokens, a scale where traditional full-attention 8B models fail because of memory constraints.

Yunyi Liu, Zhanyu Wang, Dong Xu et al.

Medical Visual Question Answering (VQA) systems play a supporting role to understand clinic-relevant information carried by medical images. The questions to a medical image include two categories: close-end (such as Yes/No question) and open-end. To obtain answers, the majority of the existing medical VQA methods relies on classification approaches, while a few works attempt to use generation approaches or a mixture of the two. The classification approaches are relatively simple but perform poorly on long open-end questions. To bridge this gap, in this paper, we propose a new Transformer based framework for medical VQA (named as Q2ATransformer), which integrates the advantages of both the classification and the generation approaches and provides a unified treatment for the close-end and open-end questions. Specifically, we introduce an additional Transformer decoder with a set of learnable candidate answer embeddings to query the existence of each answer class to a given image-question pair. Through the Transformer attention, the candidate answer embeddings interact with the fused features of the image-question pair to make the decision. In this way, despite being a classification-based approach, our method provides a mechanism to interact with the answer information for prediction like the generation-based approaches. On the other hand, by classification, we mitigate the task difficulty by reducing the search space of answers. Our method achieves new state-of-the-art performance on two medical VQA benchmarks. Especially, for the open-end questions, we achieve 79.19% on VQA-RAD and 54.85% on PathVQA, with 16.09% and 41.45% absolute improvements, respectively.

Guanbo Pan, Guo Lu, Zhihao Hu et al.

In recent years, neural image compression (NIC) algorithms have shown powerful coding performance. However, most of them are not adaptive to the image content. Although several content adaptive methods have been proposed by updating the encoder-side components, the adaptability of both latents and the decoder is not well exploited. In this work, we propose a new NIC framework that improves the content adaptability on both latents and the decoder. Specifically, to remove redundancy in the latents, our content adaptive channel dropping (CACD) method automatically selects the optimal quality levels for the latents spatially and drops the redundant channels. Additionally, we propose the content adaptive feature transformation (CAFT) method to improve decoder-side content adaptability by extracting the characteristic information of the image content, which is then used to transform the features in the decoder side. Experimental results demonstrate that our proposed methods with the encoder-side updating algorithm achieve the state-of-the-art performance.

Zicheng Wang, Zhen Zhao, Yiming Wu et al.

Unsupervised domain adaptation (UDA) is a critical challenge in the field of point cloud analysis. Previous works tackle the problem either by feature extractor adaptation to enable a shared classifier to distinguish domain-invariant features, or by classifier adaptation to evolve the classifier to recognize target-styled source features to increase its adaptation ability. However, by learning domain-invariant features, feature extractor adaptation methods fail to encode semantically meaningful target-specific information, while classifier adaptation methods rely heavily on the accurate estimation of the target distribution. In this work, we propose a novel framework that deeply couples the classifier and feature extractor adaption for 3D UDA, dubbed Progressive Classifier and Feature Extractor Adaptation (PCFEA). Our PCFEA conducts 3D UDA from two distinct perspectives: macro and micro levels. On the macro level, we propose a progressive target-styled feature augmentation (PTFA) that establishes a series of intermediate domains to enable the model to progressively adapt to the target domain. Throughout this process, the source classifier is evolved to recognize target-styled source features (\ie, classifier adaptation). On the micro level, we develop an intermediate domain feature extractor adaptation (IDFA) that performs a compact feature alignment to encourage the target-styled feature extraction gradually. In this way, PTFA and IDFA can mutually benefit each other: IDFA contributes to the distribution estimation of PTFA while PTFA constructs smoother intermediate domains to encourage an accurate feature alignment of IDFA. We validate our method on popular benchmark datasets, where our method achieves new state-of-the-art performance. Our code is available at https://github.com/xiaoyao3302/PCFEA.

Ximing Xing, Chuang Wang, Haitao Zhou et al.

Exemplar-based sketch-to-photo synthesis allows users to generate photo-realistic images based on sketches. Recently, diffusion-based methods have achieved impressive performance on image generation tasks, enabling highly-flexible control through text-driven generation or energy functions. However, generating photo-realistic images with color and texture from sketch images remains challenging for diffusion models. Sketches typically consist of only a few strokes, with most regions left blank, making it difficult for diffusion-based methods to produce photo-realistic images. In this work, we propose a two-stage method named ``Inversion-by-Inversion" for exemplar-based sketch-to-photo synthesis. This approach includes shape-enhancing inversion and full-control inversion. During the shape-enhancing inversion process, an uncolored photo is generated with the guidance of a shape-energy function. This step is essential to ensure control over the shape of the generated photo. In the full-control inversion process, we propose an appearance-energy function to control the color and texture of the final generated photo.Importantly, our Inversion-by-Inversion pipeline is training-free and can accept different types of exemplars for color and texture control. We conducted extensive experiments to evaluate our proposed method, and the results demonstrate its effectiveness. The code and project can be found at https://ximinng.github.io/inversion-by-inversion-project/.

Yandong Guan, Xilin Wang, Ximing Xing et al.

In this work, we introduce CAD-Coder, a novel framework that reformulates text-to-CAD as the generation of CadQuery scripts - a Python-based, parametric CAD language. This representation enables direct geometric validation, a richer modeling vocabulary, and seamless integration with existing LLMs. To further enhance code validity and geometric fidelity, we propose a two-stage learning pipeline: (1) supervised fine-tuning on paired text-CadQuery data, and (2) reinforcement learning with Group Reward Policy Optimization (GRPO), guided by a CAD-specific reward comprising both a geometric reward (Chamfer Distance) and a format reward. We also introduce a chain-of-thought (CoT) planning process to improve model reasoning, and construct a large-scale, high-quality dataset of 110K text-CadQuery-3D model triplets and 1.5K CoT samples via an automated pipeline. Extensive experiments demonstrate that CAD-Coder enables LLMs to generate diverse, valid, and complex CAD models directly from natural language, advancing the state of the art of text-to-CAD generation and geometric reasoning.

Qianwei Yang, Dong Xu, Zhangfan Yang et al.

Drug discovery can be viewed as a combinatorial search over an immense chemical space, motivating the development of deep generative models for de novo molecular design. Among these, GPT-based molecular language models (MLM) have shown strong molecular design performance by learning chemical syntax and semantics from large-scale data. However, existing MLMs face two fundamental limitations: they inadequately capture the graph-structured nature of molecules when formulated as next-token prediction problems, and they typically lack explicit mechanisms for target-aware generation. Here, we propose SoftMol, a unified framework that co-designs molecular representation, model architecture, and search strategy for target-aware molecular generation. SoftMol introduces soft fragments, a rule-free block representation of SMILES that enables diffusion-native modeling, and develops SoftBD, the first block-diffusion molecular language model that combines local bidirectional diffusion with autoregressive generation under molecular structural constraints. To favor generated molecules with high drug-likeness and synthetic accessibility, SoftBD is trained on a carefully curated dataset named ZINC-Curated. SoftMol further integrates a gated Monte Carlo tree search to assemble fragments in a target-aware manner. Experimental results show that, compared with current state-of-the-art models, SoftMol achieves 100% chemical validity, improves binding affinity by 9.7%, yields a 2-3x increase in molecular diversity, and delivers a 6.6x speedup in inference efficiency. Code is available at https://github.com/szu-aicourse/softmol

Fei He, Zhiyuan Yang, Mingyue Gao et al.

Cryo-electron microscopy (cryo-EM) remains pivotal in structural biology, yet the task of protein particle picking, integral for 3D protein structure construction, is laden with manual inefficiencies. While recent AI tools such as Topaz and crYOLO are advancing the field, they do not fully address the challenges of cryo-EM images, including low contrast, complex shapes, and heterogeneous conformations. This study explored prompt-based learning to adapt the state-of-the-art image segmentation foundation model Segment Anything Model (SAM) for cryo-EM. This focus was driven by the desire to optimize model performance with a small number of labeled data without altering pre-trained parameters, aiming for a balance between adaptability and foundational knowledge retention. Through trials with three prompt-based learning strategies, namely head prompt, prefix prompt, and encoder prompt, we observed enhanced performance and reduced computational requirements compared to the fine-tuning approach. This work not only highlights the potential of prompting SAM in protein identification from cryo-EM micrographs but also suggests its broader promise in biomedical image segmentation and object detection.

Dong Xu, Han Meng, Xinyu Chen et al.

Large language models (LLMs) training or inference across multiple nodes introduces significant pressure on GPU memory and interconnect bandwidth. The Compute Express Link (CXL) shared memory pool offers a scalable solution by enabling memory sharing across nodes, reducing over-provisioning and improving resource utilization. We propose \name, a collective communication library, leveraging the CXL shared memory pool to support cross-node GPU operations without relying on traditional RDMA-based networking. Our design addresses the challenges on synchronization, data interleaving, and communication parallelization faced by using the CXL shared memory pool for collective communications. Evaluating on multiple nodes with a TITAN-II CXL switch and six Micron CZ120 memory cards, we show that \name achieves highly efficient collective operations across hosts, demonstrating CXL's potential for scalable, memory-centric GPU communication. Our evaluation demonstrates that \name achieves average performance improvements of 1.34$\times$ for AllGather, 1.84$\times$ for Broadcast, 1.94$\times$ for Gather, and 1.04$\times$ for Scatter, compared to the original RDMA-based implementation over 200 Gbps InfiniBand. \textcolor{dong}{In addition, the evaluation with a case of LLM training shows 1.11$\times$ speedup compared with the InfiniBand while saving production cost by $2.75\times$ in hardware.}

Juncheng Hu, Ziteng Xue, Guotao Liang et al.

We introduce AmodalSVG, a new framework for amodal image vectorization that produces semantically organized and geometrically complete SVG representations from natural images. Existing vectorization methods operate under a modal paradigm: tracing only visible pixels and disregarding occlusion. Consequently, the resulting SVGs are semantically entangled and geometrically incomplete, limiting SVG's structural editability. In contrast, AmodalSVG reconstructs full object geometries, including occluded regions, into independent, editable vector layers. To achieve this, AmodalSVG reformulates image vectorization as a two-stage framework, performing semantic decoupling and completion in the raster domain to produce amodally complete semantic layers, which are then independently vectorized. In the first stage, we introduce Semantic Layer Peeling (SLP), a VLM-guided strategy that progressively decomposes an image into semantically coherent layers. By hybrid inpainting, SLP recovers complete object appearances under occlusions, enabling explicit semantic decoupling. To vectorize these layers efficiently, we propose Adaptive Layered Vectorization (ALV), which dynamically modulates the primitive budget via an error-budget-driven adjustment mechanism. Extensive experiments demonstrate that AmodalSVG significantly outperforms prior methods in visual fidelity. Moreover, the resulting amodal layers enable object-level editing directly in the vector domain, capabilities not supported by existing vectorization approaches. Code will be released upon acceptance.

Xinhui Liu, Zhenghao Chen, Luping Zhou et al.

Conventional Federated Domain Adaptation (FDA) approaches usually demand an abundance of assumptions, which makes them significantly less feasible for real-world situations and introduces security hazards. This paper relaxes the assumptions from previous FDAs and studies a more practical scenario named Universal Federated Domain Adaptation (UFDA). It only requires the black-box model and the label set information of each source domain, while the label sets of different source domains could be inconsistent, and the target-domain label set is totally blind. Towards a more effective solution for our newly proposed UFDA scenario, we propose a corresponding methodology called Hot-Learning with Contrastive Label Disambiguation (HCLD). It particularly tackles UFDA's domain shifts and category gaps problems by using one-hot outputs from the black-box models of various source domains. Moreover, to better distinguish the shared and unknown classes, we further present a cluster-level strategy named Mutual-Voting Decision (MVD) to extract robust consensus knowledge across peer classes from both source and target domains. Extensive experiments on three benchmark datasets demonstrate that our method achieves comparable performance for our UFDA scenario with much fewer assumptions, compared to previous methodologies with comprehensive additional assumptions.

Yuan Shui, Yandong Guan, Zhanwei Zhang et al.

Parametric Computer-Aided Design (CAD) of articulated assemblies is essential for product development, yet generating these multi-part, movable models from high-level descriptions remains unexplored. To address this, we propose ArtiCAD, the first training-free multi-agent system capable of generating editable, articulated CAD assemblies directly from text or images. Our system divides this complex task among four specialized agents: Design, Generation, Assembly, and Review. One of our key insights is to predict assembly relationships during the initial design stage rather than the assembly stage. By utilizing a Connector that explicitly defines attachment points and joint parameters, ArtiCAD determines these relationships before geometry generation, effectively bypassing the limited spatial reasoning capabilities of current LLMs and VLMs. To further ensure high-quality outputs, we introduce validation steps in the generation and assembly stages, accompanied by a cross-stage rollback mechanism that accurately isolates and corrects design- and code-level errors. Additionally, a self-evolving experience store accumulates design knowledge to continuously improve performance on future tasks. Extensive evaluations on three datasets (ArtiCAD-Bench, CADPrompt, and ACD) validate the effectiveness of our approach. We further demonstrate the applicability of ArtiCAD in requirement-driven conceptual design, physical prototyping, and the generation of embodied AI training assets through URDF export.

Zicheng Wang, Zhenghao Chen, Yiming Wu et al.

Point cloud analysis has seen substantial advancements due to deep learning, although previous Transformer-based methods excel at modeling long-range dependencies on this task, their computational demands are substantial. Conversely, the Mamba offers greater efficiency but shows limited potential compared with Transformer-based methods. In this study, we introduce PoinTramba, a pioneering hybrid framework that synergies the analytical power of Transformer with the remarkable computational efficiency of Mamba for enhanced point cloud analysis. Specifically, our approach first segments point clouds into groups, where the Transformer meticulously captures intricate intra-group dependencies and produces group embeddings, whose inter-group relationships will be simultaneously and adeptly captured by efficient Mamba architecture, ensuring comprehensive analysis. Unlike previous Mamba approaches, we introduce a bi-directional importance-aware ordering (BIO) strategy to tackle the challenges of random ordering effects. This innovative strategy intelligently reorders group embeddings based on their calculated importance scores, significantly enhancing Mamba's performance and optimizing the overall analytical process. Our framework achieves a superior balance between computational efficiency and analytical performance by seamlessly integrating these advanced techniques, marking a substantial leap forward in point cloud analysis. Extensive experiments on datasets such as ScanObjectNN, ModelNet40, and ShapeNetPart demonstrate the effectiveness of our approach, establishing a new state-of-the-art analysis benchmark on point cloud recognition. For the first time, this paradigm leverages the combined strengths of both Transformer and Mamba architectures, facilitating a new standard in the field. The code is available at https://github.com/xiaoyao3302/PoinTramba.

Jiahao Xia, Wenjian Huang, Min Xu et al.

Object parts serve as crucial intermediate representations in various downstream tasks, but part-level representation learning still has not received as much attention as other vision tasks. Previous research has established that Vision Transformer can learn instance-level attention without labels, extracting high-quality instance-level representations for boosting downstream tasks. In this paper, we achieve unsupervised part-specific attention learning using a novel paradigm and further employ the part representations to improve part discovery performance. Specifically, paired images are generated from the same image with different geometric transformations, and multiple part representations are extracted from these paired images using a novel module, named PartFormer. These part representations from the paired images are then exchanged to improve geometric transformation invariance. Subsequently, the part representations are aligned with the feature map extracted by a feature map encoder, achieving high similarity with the pixel representations of the corresponding part regions and low similarity in irrelevant regions. Finally, the geometric and semantic constraints are applied to the part representations through the intermediate results in alignment for part-specific attention learning, encouraging the PartFormer to focus locally and the part representations to explicitly include the information of the corresponding parts. Moreover, the aligned part representations can further serve as a series of reliable detectors in the testing phase, predicting pixel masks for part discovery. Extensive experiments are carried out on four widely used datasets, and our results demonstrate that the proposed method achieves competitive performance and robustness due to its part-specific attention.

Dong Xu, Jiantao Wu, Qihua Pan et al.

Drug-drug interaction (DDI) prediction is central to drug discovery and clinical development, particularly in the context of increasingly prevalent polypharmacy. Although existing computational methods achieve strong performance on standard benchmarks, they often fail to generalize to realistic deployment scenarios, where most candidate drug pairs involve previously unseen drugs and validated interactions are scarce. We demonstrate that proximity in the embedding spaces of prevailing molecule-centric DDI models does not reliably correspond to interaction labels, and that simply scaling up model capacity therefore fails to improve generalization. To address these limitations, we propose GenRel-DDI, a generalizable relation learning framework that reformulates DDI prediction as a relation-centric learning problem, in which interaction representations are learned independently of drug identities. This relation-level abstraction enables the capture of transferable interaction patterns that generalize to unseen drugs and novel drug pairs. Extensive experiments across multiple benchmark demonstrate that GenRel-DDI consistently and significantly outperforms state-of-the-art methods, with particularly large gains on strict entity-disjoint evaluations, highlighting the effectiveness and practical utility of relation learning for robust DDI prediction. The code is available at https://github.com/SZU-ADDG/GenRel-DDI.

Dong Xu, Mingwei Liu, Xiwen Wang et al.

Maintaining up-to-date, comprehensive documentation for large codebases is a persistent challenge. Recent progress in automated documentation has moved from template-based rules to large language models (LLMs), yet existing tools still process source code as flat fragments, producing isolated documents that lack semantic structure. This design also leads to excessive token consumption and slow generation, while failing to capture how code changes propagate across dependencies. We propose RepoDoc, a system that uses a repository knowledge graph (RepoKG) as the semantic foundation for the entire documentation lifecycle. Our framework consists of three stages: (1) RepoKG construction, which extracts code entities and their relationships; (2) module clustering, which groups code into functionally cohesive, hierarchical units; and (3) skillful agent-based generation, which queries the graph to create modular, cross-referenced documentation with auto-generated Mermaid diagrams. For incremental maintenance, a semantic impact propagation mechanism navigates the RepoKG bidirectionally to pinpoint all affected parts, allowing selective, targeted regeneration. Evaluated on 24 repositories across 8 programming languages, RepoDoc substantially outperforms state-of-the-art alternatives. It improves API coverage by 32.5% and completeness by 10.4%, while generating documentation 3x faster with 85% fewer tokens. For incremental updates, it cuts update time by 73% and token usage by 77%, and achieves 10.2% higher update recall, more accurately reflecting code changes in the regenerated documentation. The source code and experimental artifacts are available at https://github.com/SYSUSELab/RepoDoc.

Dong Xu, Qihua Pan, Sisi Yuan et al.

Molecular generative models, often employing GPT-style language modeling on molecular string representations, have shown promising capabilities when scaled to large datasets and model sizes. However, it remains unclear and subject to debate whether these models adhere to predictable scaling laws under fixed computational budgets, which is a crucial understanding for optimally allocating resources between model size, data volume, and molecular representation. In this study, we systematically investigate the scaling behavior of molecular language models across both pretraining and downstream tasks. We train 300 models and conduct over 10,000 experiments, rigorously controlling compute budgets while independently varying model size, number of training tokens, and molecular representation. Our results demonstrate clear scaling laws in molecular models for both pretraining and downstream transfer, reveal the substantial impact of molecular representation on performance, and explain previously observed inconsistencies in scaling behavior for molecular generation. Additionally, we publicly release the largest library of molecular language models to date to facilitate future research and development. Code and models are available at https://github.com/SZU-ADDG/MLM-Scaling.

Bowen Tang, Long Yan, Jing Zhang et al.

Deep learning models have the ability to extract rich knowledge from large-scale datasets. However, the sharing of data has become increasingly challenging due to concerns regarding data copyright and privacy. Consequently, this hampers the effective transfer of knowledge from existing data to novel downstream tasks and concepts. Zero-shot learning (ZSL) approaches aim to recognize new classes by transferring semantic knowledge learned from base classes. However, traditional generative ZSL methods often require access to real images from base classes and rely on manually annotated attributes, which presents challenges in terms of data restrictions and model scalability. To this end, this paper tackles a challenging and practical problem dubbed as data-free zero-shot learning (DFZSL), where only the CLIP-based base classes data pre-trained classifier is available for zero-shot classification. Specifically, we propose a generic framework for DFZSL, which consists of three main components. Firstly, to recover the virtual features of the base data, we model the CLIP features of base class images as samples from a von Mises-Fisher (vMF) distribution based on the pre-trained classifier. Secondly, we leverage the text features of CLIP as low-cost semantic information and propose a feature-language prompt tuning (FLPT) method to further align the virtual image features and textual features. Thirdly, we train a conditional generative model using the well-aligned virtual image features and corresponding semantic text features, enabling the generation of new classes features and achieve better zero-shot generalization. Our framework has been evaluated on five commonly used benchmarks for generalized ZSL, as well as 11 benchmarks for the base-to-new ZSL. The results demonstrate the superiority and effectiveness of our approach. Our code is available in https://github.com/ylong4/DFZSL

Guotao Liang, Zhangcheng Wang, Chuang Wang et al.

Scalable Vector Graphics (SVG) animation generation is pivotal for professional design due to their structural editability and resolution independence. However, this task remains challenging as it requires bridging discrete code representations with continuous visual dynamics. Existing optimization-based methods often destroy topological consistency, while general-purpose LLMs rely on rigid CSS/SMIL transformations, failing to model geometry-level non-rigid deformations. To address these limitations, we present VAnim, the first LLM-based framework for open-domain text-to-SVG animation. We reconceptualize animation not as sequence generation, but as Sparse State Updates (SSU) on a persistent SVG DOM tree. This paradigm compresses sequence length by over 9.8x while preserving the SVG DOM structure and non-participating elements by construction. To enable precise control, we propose an Identification-First Motion Planning mechanism that grounds textual instructions in explicit visual entities. Furthermore, to overcome the non-differentiable nature of SVG rendering, we employ Rendering-Aware Reinforcement Learning via Group Relative Policy Optimization (GRPO). By leveraging a hybrid reward from a state-of-the-art video perception encoder, we align discrete code updates with high-fidelity visual feedback. We also introduce SVGAnim-134k, the first benchmark for vector animation. Extensive experiments demonstrate that VAnim significantly outperforms state-of-the-art baselines in semantic alignment and structural validity, with additional appendix metrics further validating motion quality and identity preservation.

Luping Liu, Chao Du, Tianyu Pang et al.

The rapid advancement of text-to-image (T2I) diffusion models has enabled them to generate unprecedented results from given texts. However, as text inputs become longer, existing encoding methods like CLIP face limitations, and aligning the generated images with long texts becomes challenging. To tackle these issues, we propose LongAlign, which includes a segment-level encoding method for processing long texts and a decomposed preference optimization method for effective alignment training. For segment-level encoding, long texts are divided into multiple segments and processed separately. This method overcomes the maximum input length limits of pretrained encoding models. For preference optimization, we provide decomposed CLIP-based preference models to fine-tune diffusion models. Specifically, to utilize CLIP-based preference models for T2I alignment, we delve into their scoring mechanisms and find that the preference scores can be decomposed into two components: a text-relevant part that measures T2I alignment and a text-irrelevant part that assesses other visual aspects of human preference. Additionally, we find that the text-irrelevant part contributes to a common overfitting problem during fine-tuning. To address this, we propose a reweighting strategy that assigns different weights to these two components, thereby reducing overfitting and enhancing alignment. After fine-tuning $512 \times 512$ Stable Diffusion (SD) v1.5 for about 20 hours using our method, the fine-tuned SD outperforms stronger foundation models in T2I alignment, such as PixArt-$α$ and Kandinsky v2.2. The code is available at https://github.com/luping-liu/LongAlign.

Yiming Wu, Huan Wang, Zhenghao Chen et al.

The growing complexity of visuomotor policies poses significant challenges for deployment with heterogeneous robotic hardware constraints. However, most existing model-efficient approaches for robotic manipulation are device- and model-specific, lack generalizability, and require time-consuming per-device optimization during the adaptation process. In this work, we propose a unified framework named \textbf{D}evice-\textbf{C}onditioned \textbf{Q}uantization-\textbf{F}or-\textbf{A}ll (DC-QFA) which amortizes deployment effort with the device-conditioned quantization-aware training and hardware-constrained architecture search. Specifically, we introduce a single supernet that spans a rich design space over network architectures and mixed-precision bit-widths. It is optimized with latency- and memory-aware regularization, guided by per-device lookup tables. With this supernet, for each target platform, we can perform a once-for-all lightweight search to select an optimal subnet without any per-device re-optimization, which enables more generalizable deployment across heterogeneous hardware, and substantially reduces deployment time. To improve long-horizon stability under low precision, we further introduce multi-step on-policy distillation to mitigate error accumulation during closed-loop execution. Extensive experiments on three representative policy backbones, such as DiffusionPolicy-T, MDT-V, and OpenVLA-OFT, demonstrate that our DC-QFA achieves $2\text{-}3\times$ acceleration on edge devices, consumer-grade GPUs, and cloud platforms, with negligible performance drop in task success. Real-world evaluations on an Inovo robot equipped with a force/torque sensor further validates that our low-bit DC-QFA policies maintain stable, contact-rich manipulation even under severe quantization.

Xiawei Li, Qingyuan Xu, Jing Zhang et al.

3D point cloud semantic segmentation has a wide range of applications. Recently, weakly supervised point cloud segmentation methods have been proposed, aiming to alleviate the expensive and laborious manual annotation process by leveraging scene-level labels. However, these methods have not effectively exploited the rich geometric information (such as shape and scale) and appearance information (such as color and texture) present in RGB-D scans. Furthermore, current approaches fail to fully leverage the point affinity that can be inferred from the feature extraction network, which is crucial for learning from weak scene-level labels. Additionally, previous work overlooks the detrimental effects of the long-tailed distribution of point cloud data in weakly supervised 3D semantic segmentation. To this end, this paper proposes a simple yet effective scene-level weakly supervised point cloud segmentation method with a newly introduced multi-modality point affinity inference module. The point affinity proposed in this paper is characterized by features from multiple modalities (e.g., point cloud and RGB), and is further refined by normalizing the classifier weights to alleviate the detrimental effects of long-tailed distribution without the need of the prior of category distribution. Extensive experiments on the ScanNet and S3DIS benchmarks verify the effectiveness of our proposed method, which outperforms the state-of-the-art by ~4% to ~6% mIoU. Codes are released at https://github.com/Sunny599/AAAI24-3DWSSG-MMA.

Maomao Li, Zhen Li, Kaipeng Zhang et al.

Existing mainstream video customization methods focus on generating identity-consistent videos based on given reference images and textual prompts. Benefiting from the rapid advancement of joint audio-video generation, this paper proposes a more compelling new task: sync audio-video customization, which aims to synchronously customize both video identity and audio timbre. Specifically, given a reference image $I^{r}$ and a reference audio $A^{r}$, this novel task requires generating videos that maintain the identity of the reference image while imitating the timbre of the reference audio, with spoken content freely specifiable through user-provided textual prompts. To this end, we propose OmniCustom, a powerful DiT-based audio-video customization framework that can synthesize a video following reference image identity, audio timbre, and text prompts all at once in a zero-shot manner. Our framework is built on three key contributions. First, identity and audio timbre control are achieved through separate reference identity and audio LoRA modules that operate through self-attention layers within the base audio-video generation model. Second, we introduce a contrastive learning objective alongside the standard flow matching objective. It uses predicted flows conditioned on reference inputs as positive examples and those without reference conditions as negative examples, thereby enhancing the model ability to preserve identity and timbre. Third, we train OmniCustom on our constructed large-scale, high-quality audio-visual human dataset. Extensive experiments demonstrate that OmniCustom outperforms existing methods in generating audio-video content with consistent identity and timbre fidelity. Project page: https://omnicustom-project.github.io/page/.

Can Wang, Lei Liu, Wei Jiang et al.

Recent advances in 3D Gaussian Splatting (3DGS) have enabled significant progress in photorealistic novel view synthesis. However, traditional 3DGS relies on a slow, iterative optimization process, which limits its use in scenarios demanding real-time results. To overcome this bottleneck, recent feed-forward methods aim to predict Gaussian attributes directly from images, but they often struggle with the redundancy of Gaussian primitives and rendering quality. In this work, we introduce a transformer-based architecture specifically designed for feed-forward Gaussian Splatting. Our key insight is that spatial and semantic relationships among Gaussians can be effectively captured through a sparse attention mechanism, enabled by a Z-order strategy that organizes the unstructured Gaussian set into a spatially coherent sequence. Furthermore, we incorporate this Z-order strategy to adaptively suppress redundancy while preserving critical structural details. This allows the transformer to efficiently model context, compress Gaussian primitives, and predict Gaussian attributes in a single forward pass. Comprehensive experiments demonstrate that our method achieves fast and high-quality novel view synthesis with fewer Gaussian primitives.

Ge Yuan, Qiyuan Qiao, Jing Zhang et al.

Effective robotic manipulation requires policies that can anticipate physical outcomes and adapt to real-world environments. Effective robotic manipulation requires policies that can anticipate physical outcomes and adapt to real-world environments. In this work, we introduce a unified framework, World-Model-Driven Diffusion Policy with Online Adaptive Learning (AdaWorldPolicy) to enhance robotic manipulation under dynamic conditions with minimal human involvement. Our core insight is that world models provide strong supervision signals, enabling online adaptive learning in dynamic environments, which can be complemented by force-torque feedback to mitigate dynamic force shifts. Our AdaWorldPolicy integrates a world model, an action expert, and a force predictor-all implemented as interconnected Flow Matching Diffusion Transformers (DiT). They are interconnected via the multi-modal self-attention layers, enabling deep feature exchange for joint learning while preserving their distinct modularity characteristics. We further propose a novel Online Adaptive Learning (AdaOL) strategy that dynamically switches between an Action Generation mode and a Future Imagination mode to drive reactive updates across all three modules. This creates a powerful closed-loop mechanism that adapts to both visual and physical domain shifts with minimal overhead. Across a suite of simulated and real-robot benchmarks, our AdaWorldPolicy achieves state-of-the-art performance, with dynamical adaptive capacity to out-of-distribution scenarios.

Xinhui Liu, Can Wang, Lei Liu et al.

Free-Viewpoint Video (FVV) reconstruction enables photorealistic and interactive 3D scene visualization; however, real-time streaming is often bottlenecked by sparse-view inputs, prohibitive training costs, and bandwidth constraints. While recent 3D Gaussian Splatting (3DGS) has advanced FVV due to its superior rendering speed, Streaming Free-Viewpoint Video (SFVV) introduces additional demands for rapid optimization, high-fidelity reconstruction under sparse constraints, and minimal storage footprints. To bridge this gap, we propose StreamLoD-GS, an LoD-based Gaussian Splatting framework designed specifically for SFVV. Our approach integrates three core innovations: 1) an Anchor- and Octree-based LoD-structured 3DGS with a hierarchical Gaussian dropout technique to ensure efficient and stable optimization while maintaining high-quality rendering; 2) a GMM-based motion partitioning mechanism that separates dynamic and static content, refining dynamic regions while preserving background stability; and 3) a quantized residual refinement framework that significantly reduces storage requirements without compromising visual fidelity. Extensive experiments demonstrate that StreamLoD-GS achieves competitive or state-of-the-art performance in terms of quality, efficiency, and storage.

Daxiang Dong, Mingming Zheng, Dong Xu et al.

We present Qianfan-OCR, a 4B-parameter end-to-end vision-language model that unifies document parsing, layout analysis, and document understanding within a single architecture. It performs direct image-to-Markdown conversion and supports diverse prompt-driven tasks including table extraction, chart understanding, document QA, and key information extraction. To address the loss of explicit layout analysis in end-to-end OCR, we propose Layout-as-Thought, an optional thinking phase triggered by special think tokens that generates structured layout representations -- bounding boxes, element types, and reading order -- before producing final outputs, recovering layout grounding capabilities while improving accuracy on complex layouts. Qianfan-OCR ranks first among end-to-end models on OmniDocBench v1.5 (93.12) and OlmOCR Bench (79.8), achieves competitive results on OCRBench, CCOCR, DocVQA, and ChartQA against general VLMs of comparable scale, and attains the highest average score on public key information extraction benchmarks, surpassing Gemini-3.1-Pro, Seed-2.0, and Qwen3-VL-235B. The model is publicly accessible via the Baidu AI Cloud Qianfan platform.

Qihua Pan, Dong Xu, Jenna Xinyi Yao et al.

Powerful AI tools for drug discovery reside in isolated web apps, desktop programs, and code libraries. Such fragmentation forces scientists to manage incompatible interfaces and specialized scripts, which can be a cumbersome and repetitive process. To address this issue, a Full-pROcess druG dEsign ageNT, named FROGENT, has been proposed. Specifically, FROGENT utilizes a Large Language Model and the Model Context Protocol to integrate multiple dynamic biochemical databases, extensible tool libraries, and task-specific AI models. This agentic framework allows FROGENT to execute complicated drug discovery workflows dynamically, including component tasks such as target identification, molecule generation and retrosynthetic planning. FROGENT has been evaluated on eight benchmarks that cover various aspects of drug discovery, such as knowledge retrieval, property prediction, virtual screening, mechanistic analysis, molecular design, and synthesis. It was compared against six increasingly advanced ReAct-style agents that support code execution and literature searches. Empirical results demonstrated that FROGENT triples the best baseline performance in hit-finding and doubles it in interaction profiling, significantly outperforming both the open-source model Qwen3-32B and the commercial model GPT-4o. In addition, real-world cases have been utilized to validate the practicability and generalization of FROGENT. This development suggests that streamlining the agentic drug discovery pipeline can significantly enhance researcher productivity.

Junkai Ji, Zhangfan Yang, Dong Xu et al.

Drug discovery is a time-consuming and expensive process, with traditional high-throughput and docking-based virtual screening hampered by low success rates and limited scalability. Recent advances in generative modelling, including autoregressive, diffusion, and flow-based approaches, have enabled de novo ligand design beyond the limits of enumerative screening. Yet these models often suffer from inadequate generalization, limited interpretability, and an overemphasis on binding affinity at the expense of key pharmacological properties, thereby restricting their translational utility. Here we present Trio, a molecular generation framework integrating fragment-based molecular language modeling, reinforcement learning, and Monte Carlo tree search, for effective and interpretable closed-loop targeted molecular design. Through the three key components, Trio enables context-aware fragment assembly, enforces physicochemical and synthetic feasibility, and guides a balanced search between the exploration of novel chemotypes and the exploitation of promising intermediates within protein binding pockets. Experimental results show that Trio reliably achieves chemically valid and pharmacologically enhanced ligands, outperforming state-of-the-art approaches with improved binding affinity (+7.85%), drug-likeness (+11.10%) and synthetic accessibility (+12.05%), while expanding molecular diversity more than fourfold.

Lei Liu, Can Wang, Zhenghao Chen et al.

Recent advances in 4D Gaussian Splatting (4DGS) editing still face challenges with view, temporal, and non-editing region consistency, as well as with handling complex text instructions. To address these issues, we propose 4DGS-Craft, a consistent and interactive 4DGS editing framework. We first introduce a 4D-aware InstructPix2Pix model to ensure both view and temporal consistency. This model incorporates 4D VGGT geometry features extracted from the initial scene, enabling it to capture underlying 4D geometric structures during editing. We further enhance this model with a multi-view grid module that enforces consistency by iteratively refining multi-view input images while jointly optimizing the underlying 4D scene. Furthermore, we preserve the consistency of non-edited regions through a novel Gaussian selection mechanism, which identifies and optimizes only the Gaussians within the edited regions. Beyond consistency, facilitating user interaction is also crucial for effective 4DGS editing. Therefore, we design an LLM-based module for user intent understanding. This module employs a user instruction template to define atomic editing operations and leverages an LLM for reasoning. As a result, our framework can interpret user intent and decompose complex instructions into a logical sequence of atomic operations, enabling it to handle intricate user commands and further enhance editing performance. Compared to related works, our approach enables more consistent and controllable 4D scene editing. Our code will be made available upon acceptance.

Daxiang Dong, Mingming Zheng, Dong Xu et al.

We present Qianfan-VL, a series of multimodal large language models ranging from 3B to 70B parameters, achieving state-of-the-art performance through innovative domain enhancement techniques. Our approach employs multi-stage progressive training and high-precision data synthesis pipelines, which prove to be critical technologies for enhancing domain-specific capabilities while maintaining strong general performance. Qianfan-VL achieves comparable results to leading open-source models on general benchmarks, with state-of-the-art performance on benchmarks such as CCBench, SEEDBench IMG, ScienceQA, and MMStar. The domain enhancement strategy delivers significant advantages in OCR and document understanding, validated on both public benchmarks (OCRBench 873, DocVQA 94.75%) and in-house evaluations. Notably, Qianfan-VL-8B and 70B variants incorporate long chain-of-thought capabilities, demonstrating superior performance on mathematical reasoning (MathVista 78.6%) and logical inference tasks. All models are trained entirely on Baidu's Kunlun P800 chips, validating the capability of large-scale AI infrastructure to train SOTA-level multimodal models with over 90% scaling efficiency on 5000 chips for a single task. This work establishes an effective methodology for developing domain-enhanced multimodal models suitable for diverse enterprise deployment scenarios.