Xiangyang Zhu

Xiangyang Zhu, Renrui Zhang, Bowei He et al.

Large-scale pre-trained models have shown promising open-world performance for both vision and language tasks. However, their transferred capacity on 3D point clouds is still limited and only constrained to the classification task. In this paper, we first collaborate CLIP and GPT to be a unified 3D open-world learner, named as PointCLIP V2, which fully unleashes their potential for zero-shot 3D classification, segmentation, and detection. To better align 3D data with the pre-trained language knowledge, PointCLIP V2 contains two key designs. For the visual end, we prompt CLIP via a shape projection module to generate more realistic depth maps, narrowing the domain gap between projected point clouds with natural images. For the textual end, we prompt the GPT model to generate 3D-specific text as the input of CLIP's textual encoder. Without any training in 3D domains, our approach significantly surpasses PointCLIP by +42.90%, +40.44%, and +28.75% accuracy on three datasets for zero-shot 3D classification. On top of that, V2 can be extended to few-shot 3D classification, zero-shot 3D part segmentation, and 3D object detection in a simple manner, demonstrating our generalization ability for unified 3D open-world learning.

Xiangyang Zhu, Renrui Zhang, Bowei He et al.

The popularity of Contrastive Language-Image Pre-training (CLIP) has propelled its application to diverse downstream vision tasks. To improve its capacity on downstream tasks, few-shot learning has become a widely-adopted technique. However, existing methods either exhibit limited performance or suffer from excessive learnable parameters. In this paper, we propose APE, an Adaptive Prior rEfinement method for CLIP's pre-trained knowledge, which achieves superior accuracy with high computational efficiency. Via a prior refinement module, we analyze the inter-class disparity in the downstream data and decouple the domain-specific knowledge from the CLIP-extracted cache model. On top of that, we introduce two model variants, a training-free APE and a training-required APE-T. We explore the trilateral affinities between the test image, prior cache model, and textual representations, and only enable a lightweight category-residual module to be trained. For the average accuracy over 11 benchmarks, both APE and APE-T attain state-of-the-art and respectively outperform the second-best by +1.59% and +1.99% under 16 shots with x30 less learnable parameters.

Shihan Ma, Alexander Kenneth Clarke, Kostiantyn Maksymenko et al.

Numerical models of electromyographic (EMG) signals have provided a huge contribution to our fundamental understanding of human neurophysiology and remain a central pillar of motor neuroscience and the development of human-machine interfaces. However, whilst modern biophysical simulations based on finite element methods are highly accurate, they are extremely computationally expensive and thus are generally limited to modelling static systems such as isometrically contracting limbs. As a solution to this problem, we propose a transfer learning approach, in which a conditional generative model is trained to mimic the output of an advanced numerical model. To this end, we present BioMime, a conditional generative neural network trained adversarially to generate motor unit activation potential waveforms under a wide variety of volume conductor parameters. We demonstrate the ability of such a model to predictively interpolate between a much smaller number of numerical model's outputs with a high accuracy. Consequently, the computational load is dramatically reduced, which allows the rapid simulation of EMG signals during truly dynamic and naturalistic movements.

Xiangyang Zhu, Yiling Pan, Bailin Deng et al.

Recovering the shape and appearance of real-world objects from natural 2D images is a long-standing and challenging inverse rendering problem. In this paper, we introduce a novel hybrid differentiable rendering method to efficiently reconstruct the 3D geometry and reflectance of a scene from multi-view images captured by conventional hand-held cameras. Our method follows an analysis-by-synthesis approach and consists of two phases. In the initialization phase, we use traditional SfM and MVS methods to reconstruct a virtual scene roughly matching the real scene. Then in the optimization phase, we adopt a hybrid approach to refine the geometry and reflectance, where the geometry is first optimized using an approximate differentiable rendering method, and the reflectance is optimized afterward using a physically-based differentiable rendering method. Our hybrid approach combines the efficiency of approximate methods with the high-quality results of physically-based methods. Extensive experiments on synthetic and real data demonstrate that our method can produce reconstructions with similar or higher quality than state-of-the-art methods while being more efficient.

Qi Qin, Le Zhuo, Yi Xin et al.

We introduce Lumina-Image 2.0, an advanced text-to-image generation framework that achieves significant progress compared to previous work, Lumina-Next. Lumina-Image 2.0 is built upon two key principles: (1) Unification - it adopts a unified architecture (Unified Next-DiT) that treats text and image tokens as a joint sequence, enabling natural cross-modal interactions and allowing seamless task expansion. Besides, since high-quality captioners can provide semantically well-aligned text-image training pairs, we introduce a unified captioning system, Unified Captioner (UniCap), specifically designed for T2I generation tasks. UniCap excels at generating comprehensive and accurate captions, accelerating convergence and enhancing prompt adherence. (2) Efficiency - to improve the efficiency of our proposed model, we develop multi-stage progressive training strategies and introduce inference acceleration techniques without compromising image quality. Extensive evaluations on academic benchmarks and public text-to-image arenas show that Lumina-Image 2.0 delivers strong performances even with only 2.6B parameters, highlighting its scalability and design efficiency. We have released our training details, code, and models at https://github.com/Alpha-VLLM/Lumina-Image-2.0.

Kaiwei Zhang, Qi Jia, Zijian Chen et al.

Large language models (LLMs), while increasingly used in domains requiring factual rigor, often display a troubling behavior: sycophancy, the tendency to align with user beliefs regardless of correctness. This tendency is reinforced by preference-based alignment techniques that optimize for user satisfaction but can undermine truthfulness. While relatively benign in casual dialogue, sycophancy poses serious risks in high-stakes settings such as scientific question answering (QA), where model outputs may shape collaborative reasoning, decision-making, and knowledge formation. Despite its importance, this phenomenon remains underexamined in factual QA contexts. We address this gap by introducing a unified evaluation framework to quantify the impact of sycophantic context on model behavior in scientific QA, measuring how much user-imposed social pressure distorts model outputs. The framework incorporates adversarial prompting setups and targeted metrics, such as misleading resistance and sycophancy resistance, that capture a model's ability to maintain factual consistency under misleading cues. Systematic evaluations across open-source and proprietary models reveal pervasive sycophantic tendencies, driven more by alignment strategy than by model size. To mitigate this issue, we propose Pressure-Tune, a lightweight post-training method that fine-tunes models on synthetic adversarial dialogues paired with chain-of-thought rationales. These rationales reject user misinformation while reinforcing factual commitments. Experiments on challenging scientific QA benchmarks show that Pressure-Tune significantly enhances sycophancy resistance without compromising accuracy or responsiveness to valid feedback, offering a practical pathway toward more truthful and principled model behavior.

Wei Sun, Weixia Zhang, Linhan Cao et al.

Face image quality assessment (FIQA) is essential for various face-related applications. Although FIQA has been extensively studied and achieved significant progress, the computational complexity of FIQA algorithms remains a key concern for ensuring scalability and practical deployment in real-world systems. In this paper, we aim to develop a computationally efficient FIQA method that can be easily deployed in real-world applications. Specifically, our method consists of two stages: training a powerful teacher model and distilling a lightweight student model from it. To build a strong teacher model, we adopt a self-training strategy to improve its capacity. We first train the teacher model using labeled face images, then use it to generate pseudo-labels for a set of unlabeled images. These pseudo-labeled samples are used in two ways: (1) to distill knowledge into the student model, and (2) to combine with the original labeled images to further enhance the teacher model through self-training. The enhanced teacher model is used to further pseudo-label another set of unlabeled images for distilling the student models. The student model is trained using a combination of labeled images, pseudo-labeled images from the original teacher model, and pseudo-labeled images from the enhanced teacher model. Experimental results demonstrate that our student model achieves comparable performance to the teacher model with an extremely low computational overhead. Moreover, our method achieved first place in the ICCV 2025 VQualA FIQA Challenge. The code is available at https://github.com/sunwei925/Efficient-FIQA.git.

Hanqing Wang, Yuan Tian, Mingyu Liu et al.

In the rapidly evolving landscape of Multimodal Large Language Models (MLLMs), the safety concerns of their outputs have earned significant attention. Although numerous datasets have been proposed, they may become outdated with MLLM advancements and are susceptible to data contamination issues. To address these problems, we propose \textbf{SDEval}, the \textit{first} safety dynamic evaluation framework to controllably adjust the distribution and complexity of safety benchmarks. Specifically, SDEval mainly adopts three dynamic strategies: text, image, and text-image dynamics to generate new samples from original benchmarks. We first explore the individual effects of text and image dynamics on model safety. Then, we find that injecting text dynamics into images can further impact safety, and conversely, injecting image dynamics into text also leads to safety risks. SDEval is general enough to be applied to various existing safety and even capability benchmarks. Experiments across safety benchmarks, MLLMGuard and VLSBench, and capability benchmarks, MMBench and MMVet, show that SDEval significantly influences safety evaluation, mitigates data contamination, and exposes safety limitations of MLLMs. Code is available at https://github.com/hq-King/SDEval

Ziyu Guo, Renrui Zhang, Xiangyang Zhu et al.

We introduce Point-Bind, a 3D multi-modality model aligning point clouds with 2D image, language, audio, and video. Guided by ImageBind, we construct a joint embedding space between 3D and multi-modalities, enabling many promising applications, e.g., any-to-3D generation, 3D embedding arithmetic, and 3D open-world understanding. On top of this, we further present Point-LLM, the first 3D large language model (LLM) following 3D multi-modal instructions. By parameter-efficient fine-tuning techniques, Point-LLM injects the semantics of Point-Bind into pre-trained LLMs, e.g., LLaMA, which requires no 3D instruction data, but exhibits superior 3D and multi-modal question-answering capacity. We hope our work may cast a light on the community for extending 3D point clouds to multi-modality applications. Code is available at https://github.com/ZiyuGuo99/Point-Bind_Point-LLM.

Xiangyang Zhu, Renrui Zhang, Bowei He et al.

To reduce the reliance on large-scale datasets, recent works in 3D segmentation resort to few-shot learning. Current 3D few-shot segmentation methods first pre-train models on 'seen' classes, and then evaluate their generalization performance on 'unseen' classes. However, the prior pre-training stage not only introduces excessive time overhead but also incurs a significant domain gap on 'unseen' classes. To tackle these issues, we propose a Non-parametric Network for few-shot 3D Segmentation, Seg-NN, and its Parametric variant, Seg-PN. Without training, Seg-NN extracts dense representations by hand-crafted filters and achieves comparable performance to existing parametric models. Due to the elimination of pre-training, Seg-NN can alleviate the domain gap issue and save a substantial amount of time. Based on Seg-NN, Seg-PN only requires training a lightweight QUEry-Support Transferring (QUEST) module, which enhances the interaction between the support set and query set. Experiments suggest that Seg-PN outperforms previous state-of-the-art method by +4.19% and +7.71% mIoU on S3DIS and ScanNet datasets respectively, while reducing training time by -90%, indicating its effectiveness and efficiency.

Linhan Cao, Wei Sun, Xiangyang Zhu et al.

Video quality assessment (VQA) seeks to predict the perceptual quality of a video in alignment with human visual perception, serving as a fundamental tool for quantifying quality degradation across video processing workflows. The dominant VQA paradigm relies on supervised training with human-labeled datasets, which, despite substantial progress, still suffers from poor generalization to unseen video content. Moreover, its reliance on human annotations -- which are labor-intensive and costly -- makes it difficult to scale datasets for improving model generalization. In this work, we explore weak-to-strong (W2S) learning as a new paradigm for advancing VQA without reliance on large-scale human-labeled datasets. We first provide empirical evidence that a straightforward W2S strategy allows a strong student model to not only match its weak teacher on in-domain benchmarks but also surpass it on out-of-distribution (OOD) benchmarks, revealing a distinct weak-to-strong effect in VQA. Building on this insight, we propose a novel framework that enhances W2S learning from two aspects: (1) integrating homogeneous and heterogeneous supervision signals from diverse VQA teachers -- including off-the-shelf VQA models and synthetic distortion simulators -- via a learn-to-rank formulation, and (2) iterative W2S training, where each strong student is recycled as the teacher in subsequent cycles, progressively focusing on challenging cases. Extensive experiments show that our method achieves state-of-the-art results across both in-domain and OOD benchmarks, with especially strong gains in OOD scenarios. Our findings highlight W2S learning as a principled route to break annotation barriers and achieve scalable generalization in VQA, with implications extending to broader alignment and evaluation tasks.

Linhan Cao, Wei Sun, Weixia Zhang et al.

Video quality assessment (VQA) aims to objectively quantify perceptual quality degradation in alignment with human visual perception. Despite recent advances, existing VQA models still suffer from two critical limitations: \textit{poor generalization to out-of-distribution (OOD) videos} and \textit{limited explainability}, which restrict their applicability in real-world scenarios. To address these challenges, we propose \textbf{VQAThinker}, a reasoning-based VQA framework that leverages large multimodal models (LMMs) with reinforcement learning to jointly model video quality understanding and scoring, emulating human perceptual decision-making. Specifically, we adopt group relative policy optimization (GRPO), a rule-guided reinforcement learning algorithm that enables reasoning over video quality under score-level supervision, and introduce three VQA-specific rewards: (1) a \textbf{bell-shaped regression reward} that increases rapidly as the prediction error decreases and becomes progressively less sensitive near the ground truth; (2) a \textbf{pairwise ranking reward} that guides the model to correctly determine the relative quality between video pairs; and (3) a \textbf{temporal consistency reward} that encourages the model to prefer temporally coherent videos over their perturbed counterparts. Extensive experiments demonstrate that VQAThinker achieves state-of-the-art performance on both in-domain and OOD VQA benchmarks, showing strong generalization for video quality scoring. Furthermore, evaluations on video quality understanding tasks validate its superiority in distortion attribution and quality description compared to existing explainable VQA models and LMMs. These findings demonstrate that reinforcement learning offers an effective pathway toward building generalizable and explainable VQA models solely with score-level supervision.

Xiangyang Zhu, Yuan Tian, Chunyi Li et al.

The rapid proliferation of large language models (LLMs) has intensified the requirement for reliable safety evaluation to uncover model vulnerabilities. To this end, numerous LLM safety evaluation benchmarks are proposed. However, existing benchmarks generally rely on labor-intensive manual curation, which causes excessive time and resource consumption. They also exhibit significant redundancy and limited difficulty. To alleviate these problems, we introduce SafetyFlow, the first agent-flow system designed to automate the construction of LLM safety benchmarks. SafetyFlow can automatically build a comprehensive safety benchmark in only four days without any human intervention by orchestrating seven specialized agents, significantly reducing time and resource cost. Equipped with versatile tools, the agents of SafetyFlow ensure process and cost controllability while integrating human expertise into the automatic pipeline. The final constructed dataset, SafetyFlowBench, contains 23,446 queries with low redundancy and strong discriminative power. Our contribution includes the first fully automated benchmarking pipeline and a comprehensive safety benchmark. We evaluate the safety of 49 advanced LLMs on our dataset and conduct extensive experiments to validate our efficacy and efficiency.

Le Zhuo, Ruoyi Du, Han Xiao et al.

Lumina-T2X is a nascent family of Flow-based Large Diffusion Transformers that establishes a unified framework for transforming noise into various modalities, such as images and videos, conditioned on text instructions. Despite its promising capabilities, Lumina-T2X still encounters challenges including training instability, slow inference, and extrapolation artifacts. In this paper, we present Lumina-Next, an improved version of Lumina-T2X, showcasing stronger generation performance with increased training and inference efficiency. We begin with a comprehensive analysis of the Flag-DiT architecture and identify several suboptimal components, which we address by introducing the Next-DiT architecture with 3D RoPE and sandwich normalizations. To enable better resolution extrapolation, we thoroughly compare different context extrapolation methods applied to text-to-image generation with 3D RoPE, and propose Frequency- and Time-Aware Scaled RoPE tailored for diffusion transformers. Additionally, we introduced a sigmoid time discretization schedule to reduce sampling steps in solving the Flow ODE and the Context Drop method to merge redundant visual tokens for faster network evaluation, effectively boosting the overall sampling speed. Thanks to these improvements, Lumina-Next not only improves the quality and efficiency of basic text-to-image generation but also demonstrates superior resolution extrapolation capabilities and multilingual generation using decoder-based LLMs as the text encoder, all in a zero-shot manner. To further validate Lumina-Next as a versatile generative framework, we instantiate it on diverse tasks including visual recognition, multi-view, audio, music, and point cloud generation, showcasing strong performance across these domains. By releasing all codes and model weights, we aim to advance the development of next-generation generative AI capable of universal modeling.

Xiangyang Zhu, Renrui Zhang, Bowei He et al.

To reduce the reliance on large-scale datasets, recent works in 3D segmentation resort to few-shot learning. Current 3D few-shot semantic segmentation methods first pre-train the models on `seen' classes, and then evaluate their generalization performance on `unseen' classes. However, the prior pre-training stage not only introduces excessive time overhead, but also incurs a significant domain gap on `unseen' classes. To tackle these issues, we propose an efficient Training-free Few-shot 3D Segmentation netwrok, TFS3D, and a further training-based variant, TFS3D-T. Without any learnable parameters, TFS3D extracts dense representations by trigonometric positional encodings, and achieves comparable performance to previous training-based methods. Due to the elimination of pre-training, TFS3D can alleviate the domain gap issue and save a substantial amount of time. Building upon TFS3D, TFS3D-T only requires to train a lightweight query-support transferring attention (QUEST), which enhances the interaction between the few-shot query and support data. Experiments demonstrate TFS3D-T improves previous state-of-the-art methods by +6.93% and +17.96% mIoU respectively on S3DIS and ScanNet, while reducing the training time by -90%, indicating superior effectiveness and efficiency.

Peng Peng, Wei Dong, Gang Chen et al.

This paper proposes a perception-shared and swarm trajectory global optimal (STGO) algorithm fused UAVs formation motion planning framework aided by an active sensing system. First, the point cloud received by each UAV is fit by the gaussian mixture model (GMM) and transmitted in the swarm. Resampling from the received GMM contributes to a global map, which is used as the foundation for consensus. Second, to improve flight safety, an active sensing system is designed to plan the observation angle of each UAV considering the unknown field, overlap of the field of view (FOV), velocity direction and smoothness of yaw rotation, and this planning problem is solved by the distributed particle swarm optimization (DPSO) algorithm. Last, for the formation motion planning, to ensure obstacle avoidance, the formation structure is allowed for affine transformation and is treated as the soft constraint on the control points of the B-spline. Besides, the STGO is introduced to avoid local minima. The combination of GMM communication and STGO guarantees a safe and strict consensus between UAVs. Tests on different formations in the simulation show that our algorithm can contribute to a strict consensus and has a success rate of at least 80% for obstacle avoidance in a dense environment. Besides, the active sensing system can increase the success rate of obstacle avoidance from 50% to 100% in some scenarios.

Gang Chen, Wei Dong, Peng Peng et al.

Particle-based dynamic occupancy maps were proposed in recent years to model the obstacles in dynamic environments. Current particle-based maps describe the occupancy status in discrete grid form and suffer from the grid size problem, wherein a large grid size is unfavorable for motion planning, while a small grid size lowers efficiency and causes gaps and inconsistencies. To tackle this problem, this paper generalizes the particle-based map into continuous space and builds an efficient 3D egocentric local map. A dual-structure subspace division paradigm, composed of a voxel subspace division and a novel pyramid-like subspace division, is proposed to propagate particles and update the map efficiently with the consideration of occlusions. The occupancy status of an arbitrary point in the map space can then be estimated with the particles' weights. To further enhance the performance of simultaneously modeling static and dynamic obstacles and minimize noise, an initial velocity estimation approach and a mixture model are utilized. Experimental results show that our map can effectively and efficiently model both dynamic obstacles and static obstacles. Compared to the state-of-the-art grid-form particle-based map, our map enables continuous occupancy estimation and substantially improves the performance in different resolutions.

Ziying Lin, Wei Dong, Sensen Liu et al.

Flying robots such as the quadrotor could provide an efficient approach for medical treatment or sensor placing of wild animals. In these applications, continuously targeting the moving animal is a crucial requirement. Due to the underactuated characteristics of the quadrotor and the coupled kinematics with the animal, nonlinear optimal tracking approaches, other than smooth feedback control, are required. However, with severe nonlinearities, it would be time-consuming to evaluate control inputs, and real-time tracking may not be achieved with generic optimizers onboard. To tackle this problem, a novel efficient egocentric regulation approach with high computational efficiency is proposed in this paper. Specifically, it directly formulates the optimal tracking problem in an egocentric manner regarding the quadrotor's body coordinates. Meanwhile, the nonlinearities of the system are peeled off through a mapping of the feedback states as well as control inputs, between the inertial and body coordinates. In this way, the proposed efficient egocentric regulator only requires solving a quadratic performance objective with linear constraints and then generate control inputs analytically. Comparative simulations and mimic biological experiment are carried out to verify the effectiveness and computational efficiency. Results demonstrate that the proposed control approach presents the highest and stablest computational efficiency than generic optimizers on different platforms. Particularly, on a commonly utilized onboard computer, our method can compute the control action in approximately 0.3 ms, which is on the order of 350 times faster than that of generic nonlinear optimizers, establishing a control frequency around 3000 Hz.

Gang Chen, Wei Dong, Xinjun Sheng et al.

This paper investigates a novel active-sensing-based obstacle avoidance paradigm for flying robots in dynamic environments. Instead of fusing multiple sensors to enlarge the field of view (FOV), we introduce an alternative approach that utilizes a stereo camera with an independent rotational DOF to sense the obstacles actively. In particular, the sensing direction is planned heuristically by multiple objectives, including tracking dynamic obstacles, observing the heading direction, and exploring the previously unseen area. With the sensing result, a flight path is then planned based on real-time sampling and uncertainty-aware collision checking in the state space, which constitutes an active sense and avoid (ASAA) system. Experiments in both simulation and the real world demonstrate that this system can well cope with dynamic obstacles and abrupt goal direction changes. Since only one stereo camera is utilized, this system provides a low-cost and effective approach to overcome the FOV limitation in visual navigation.

Gang Chen, Hongzhe Yu, Wei Dong et al.

While training an end-to-end navigation network in the real world is usually of high cost, simulation provides a safe and cheap environment in this training stage. However, training neural network models in simulation brings up the problem of how to effectively transfer the model from simulation to the real world (sim-to-real). In this work, we regard the environment representation as a crucial element in this transfer process and propose a visual information pyramid (VIP) model to systematically investigate a practical environment representation. A novel representation composed of spatial and semantic information synthesis is then established accordingly, where noise model embedding is particularly considered. To explore the effectiveness of this representation, we compared the performance with representations popularly used in the literature in both simulated and real-world scenarios. Results suggest that our environment representation stands out. Furthermore, an analysis on the feature map is implemented to investigate the effectiveness through inner reaction, which could be irradiative for future researches on end-to-end navigation.