论文检索

Jiankang Wang, Zhihan Zhang, Zhihang Liu, Yang Li, Jiannan Ge, Hongtao Xie, Yongdong Zhang

Multimodal Large Language Models (MLLMs) have shown remarkable progress in temporal or spatial localization tasks, but struggle with joint spatio-temporal video grounding (STVG). We identify two key bottlenecks hindering this capability: (1) the sheer number of visual tokens makes long-range and fine-grained visual modeling challenging; (2) generating a long sequence of bounding boxes in text makes it hard to accurately align each box with its specific video frame. Distinct from prior efforts that rely on attaching complex modules, we argue for a more elegant paradigm that unlocks the inherent potential of MLLMs and leverages their strengths. To this end, we propose \textbf{\textit{SpaceVLLM}}, a MLLM equipped with spatio-temporal video grounding capabilities. Specifically, we propose Spatio-Temporal Aware Queries, interleaved with video frames, to guide the MLLM in capturing both static appearance and dynamic motion features. We further present a lightweight Query-Guided Space Head that maps queries to precise spatial coordinates, bypassing the need for direct textual coordinate generation and enabling the MLLM to focus on video understanding. To further facilitate research in this area, we propose an automated data synthesis pipeline to construct \textbf{V-STG} dataset, comprising 110K STVG instances. Extensive experiments show that \textit{SpaceVLLM} achieves the state-of-the-art performance on STVG benchmarks and maintains strong performance on various video understanding tasks, validating our approach's effectiveness.

Jian Wang, Lijun He, Yixing Yong, Haixia Bi, Fan Li

Modern autonomous driving (AD) systems leverage 3D object detection to perceive foreground objects in 3D environments for subsequent prediction and planning. Visual 3D detection based on RGB cameras provides a cost-effective solution compared to the LiDAR paradigm. While achieving promising detection accuracy, current deep neural network-based models remain highly susceptible to adversarial examples. The underlying safety concerns motivate us to investigate realistic adversarial attacks in AD scenarios. Previous work has demonstrated the feasibility of placing adversarial posters on the road surface to induce hallucinations in the detector. However, the unnatural appearance of the posters makes them easily noticeable by humans, and their fixed content can be readily targeted and defended. To address these limitations, we propose the AdvRoad to generate diverse road-style adversarial posters. The adversaries have naturalistic appearances resembling the road surface while compromising the detector to perceive non-existent objects at the attack locations. We employ a two-stage approach, termed Road-Style Adversary Generation and Scenario-Associated Adaptation, to maximize the attack effectiveness on the input scene while ensuring the natural appearance of the poster, allowing the attack to be carried out stealthily without drawing human attention. Extensive experiments show that AdvRoad generalizes well to different detectors, scenes, and spoofing locations. Moreover, physical attacks further demonstrate the practical threats in real-world environments.

Jiale Wang, Chen Zhao, Wei Ke, Tong Zhang

Random Sample Consensus (RANSAC) is a fundamental approach for robustly estimating parametric models from noisy data. Existing learning-based RANSAC methods utilize deep learning to enhance the robustness of RANSAC against outliers. However, these approaches are trained and tested on the data generated by the same algorithms, leading to limited generalization to out-of-distribution data during inference. Therefore, in this paper, we introduce a novel diffusion-based paradigm that progressively injects noise into ground-truth data, simulating the noisy conditions for training learning-based RANSAC. To enhance data diversity, we incorporate Monte Carlo sampling into the diffusion paradigm, approximating diverse data distributions by introducing different types of randomness at multiple stages. We evaluate our approach in the context of feature matching through comprehensive experiments on the ScanNet and MegaDepth datasets. The experimental results demonstrate that our Monte Carlo diffusion mechanism significantly improves the generalization ability of learning-based RANSAC. We also develop extensive ablation studies that highlight the effectiveness of key components in our framework.

Jiahui Wang, Haiyue Zhu, Haoren Guo, Abdullah Al Mamun, Cheng Xiang, Tong Heng Lee

Recent approaches for few-shot 3D point cloud semantic segmentation typically require a two-stage learning process, i.e., a pre-training stage followed by a few-shot training stage. While effective, these methods face overreliance on pre-training, which hinders model flexibility and adaptability. Some models tried to avoid pre-training yet failed to capture ample information. In addition, current approaches focus on visual information in the support set and neglect or do not fully exploit other useful data, such as textual annotations. This inadequate utilization of support information impairs the performance of the model and restricts its zero-shot ability. To address these limitations, we present a novel pre-training-free network, named Efficient Point Cloud Semantic Segmentation for Few- and Zero-shot scenarios. Our EPSegFZ incorporates three key components. A Prototype-Enhanced Registers Attention (ProERA) module and a Dual Relative Positional Encoding (DRPE)-based cross-attention mechanism for improved feature extraction and accurate query-prototype correspondence construction without pre-training. A Language-Guided Prototype Embedding (LGPE) module that effectively leverages textual information from the support set to improve few-shot performance and enable zero-shot inference.Extensive experiments show that our method outperforms the state-of-the-art method by 5.68% and 3.82% on the S3DIS and ScanNet benchmarks, respectively.

Jiahao Wang, Zhongwei Jiang, Wenchao Sun, Jiaru Zhong, Haibao Yu, Yuner Zhang, Chenyang Lu, Chuang Zhang, Lei He, Shaobing Xu 等

Cooperative perception is critical for autonomous driving, overcoming the inherent limitations of a single vehicle, such as occlusions and constrained fields-of-view. However, current approaches sharing dense Bird's-Eye-View (BEV) features are constrained by quadratically-scaling communication costs and the lack of flexibility and interpretability for precise alignment across asynchronous or disparate viewpoints. While emerging sparse query-based methods offer an alternative, they often suffer from inadequate geometric representations, suboptimal fusion strategies, and training instability. In this paper, we propose SparseCoop, a fully sparse cooperative perception framework for 3D detection and tracking that completely discards intermediate BEV representations. Our framework features a trio of innovations: a kinematic grounded instance query that uses an explicit state vector with 3D geometry and velocity for precise spatio-temporal alignment; a coarse-to-fine aggregation module that effectively integrates information from both matched and unmatched instances; and a cooperative instance denoising task that provides stable, abundant supervision to accelerate and stabilize training. Experiments on V2X-Seq and Griffin datasets show SparseCoop achieves state-of-the-art performance. Notably, it delivers this performance with superior computational efficiency and a highly competitive transmission cost, while showing remarkable robustness to real-world challenges like communication latency.

Jiahao Wang, Xiangyu Cao, Jiaru Zhong, Yuner Zhang, Zeyu Han, Haibao Yu, Chuang Zhang, Lei He, Shaobing Xu, Jianqiang Wang

While cooperative perception can overcome the limitations of single-vehicle systems, the practical implementation of vehicle-to-vehicle and vehicle-to-infrastructure systems is often impeded by significant economic barriers. Aerial-ground cooperation (AGC), which pairs ground vehicles with drones, presents a more economically viable and rapidly deployable alternative. However, this emerging field has been held back by a critical lack of high-quality public datasets and benchmarks. To bridge this gap, we present Griffin, a comprehensive AGC 3D perception dataset, featuring over 250 dynamic scenes (37k+ frames). It incorporates varied drone altitudes (20-60m), diverse weather conditions, realistic drone dynamics via CARLA-AirSim co-simulation, and critical occlusion-aware 3D annotations. Accompanying the dataset is a unified benchmarking framework for cooperative detection and tracking, with protocols to evaluate communication efficiency, altitude adaptability, and robustness to communication latency, data loss and localization noise. By experiments through different cooperative paradigms, we demonstrate the effectiveness and limitations of current methods and provide crucial insights for future research.

Jiahao Wang, Fang Liu, Licheng Jiao, Hao Wang, Shuo Li, Xinyi Wang, Lingling Li, Puhua Chen, Xu Liu

In recent years, the rapid progress of deep learning has driven notable advancements in satellite video tracking, a critical task for applications such as environmental monitoring, disaster management, and defense. Despite these strides, existing approaches remain constrained by their inability to handle dynamic challenges, such as target appearance variations, complex motion patterns, and occlusions. Traditional methods often suffer from static template matching or overly complex update mechanisms, compromising their robustness and practicality in real-world scenarios. To address these limitations, we propose a paradigm shift in satellite video tracking by integrating historical trajectory knowledge with visual features. This fusion enhances the tracker's perceptual understanding of targets over time, enabling more adaptive and resilient tracking. By aligning spatial, temporal, and cross-modal information, our approach effectively bridges the gap between fragmented observations and coherent tracking performance, even under challenging conditions like small target detection and cluttered backgrounds. Extensive experiments conducted on multiple satellite video tracking benchmarks demonstrate the superiority of our method, with HTTrack achieving success rates of 51.5% on SV248S, 52.9% on SatSOT, and 32.6% on VISO, significantly outperforming state-of-the-art trackers and marking a step forward in achieving robust, accurate, and scalable satellite video tracking.

Jiahao Wang, Fang Liu, Hao Wang, Shuo Li, Xinyi Wang, Puhua Chen

RGB-T tracking is increasingly deployed in safety-critical applications such as autonomous driving, surveillance, and rescue robotics, where tracking reliability is essential under adverse conditions. Although the fusion of RGB and thermal infrared (TIR) modalities offers improved robustness in low-light and occluded scenes, recent findings show that RGB-T trackers remain highly susceptible to subtle input perturbations, human-imperceptible modifications that exploit cross-modal inconsistencies to mislead tracking outputs. In real-world scenarios, such perturbations can arise from sensor spoofing, infrared camouflage, or physical-world attacks, posing serious risks to operational safety. To address this, we propose SFPT, a Semantic Feature Purification framework that enhances RGB-T tracking at the representation level. Rather than filtering corrupted inputs at the pixel level, SFPT introduces task-specific semantic anchors into the feature space to reinforce perturbation-invariant cues. These anchors are derived from descriptive language, interact with visual features to purify representations. To further suppress modality-specific interference, we design an Adaptive Perturbation-Guided Cross-Modal Fusion (APG-CMF) module, which leverages language and visual signals to estimate reliability and dynamically reweight cross-modal features, ensuring robust fusion under perturbation conditions. Extensive experiments under diverse perturbation conditions validate the effectiveness of our approach. Notably, SFPT maintains performance comparable to clean settings even when subjected to perturbations of strength 1/255 and 4/255, demonstrating strong resilience to real-world interference.

Huan Wang, Haoran Li, Yuxin Lin, Huaming Chen, Jun Yan, Lijuan Wang, Jiahua Shi, Qihao Xu, Yongting Hu, Yong Xu 等

As one of the primary causes of visual impairment, Diabetic Retinopathy (DR) requires accurate and robust grading to facilitate timely diagnosis and intervention. Different from conventional DR grading methods that utilize single-view images, recent clinical studies have revealed that multi-view fundus images can significantly enhance DR grading performance by expanding the field of view (FOV). However, there is a long-tailed distribution problem in fundus image analysis, i.e., a high prevalence of mild DR grades and a low prevalence of rare ones (e.g., cases of high severity), which presents a significant challenge to developing a unified model capable of detecting rare or unseen DR grades not encountered during training. In this paper, we propose ProME-DR, a Prompt-driven zero-shot DR grading framework, which leverages prompt Matching and Emulating to recognize the unseen DR categories and views beyond the training set. ProME-DR disentangles the training process into two stages to learn generalized knowledge for novel DR disease grading. Initially, ProME-DR leverages two sets of prompt units to capture semantic and inter-view consistency knowledge via a split-and-mask manner, gathering instance-level DR visual clues. Subsequently, it constructs a concept-aware emulator to generate context prompt units, linking extensible knowledge learned from the previously seen DR attributes for zero-shot DR grading. Extensive experiments conducted on eight datasets and various scenarios confirm the superiority of ProME-DR.

Huabin Wang, Yu Yang, Xinran Zhong, Zilong Ling

Transmitting and receiving electromagnetic wave signals reflected back to the ground can detect the structure of subsurface defects. However, the imaging process of ground-penetrating radar (GPR) is highly susceptible to interference from complex underground environments, leading to nonlinear attenuation and noise. This makes it challenging to directly locate and identify defect types from raw reflected radar waveform images. Currently, mainstream methods of manual radar signal gain and filtering heavily rely on expert experience, while common end-to-end generative models are typically designed for optical images. This paper proposes a defect-guided Multi-window Gabor Transform Network (MGT-Net) for GPR B-Scan image reconstruction which achieves automatic gain and defect enhancement of raw GPR signals. Firstly, a Multi-window Gabor Transform Module (MGTM) is designed to effectively represent and extract spatial-frequency features of defects at different locations and of various types. Secondly, a defect guidance network (DG-Net) is constructed to accurately direct the reconstruction of defect areas and enhance the saliency and discriminability of defect features. Additionally, we construct a large-scale GPR B-Scan image dataset (GRD) containing 41,613 images across 7 defect categories. Experimental results show the superior performance of MGT-Net, achieving state-of-the-art (SOTA) SSIM of 81.72% ± 3.5% and PSNR of 30.50 ± 0.442.

Hua Wang, Xiaodan Zhang, Yanzhao Shi, Chengxin Zheng, Wanyu Zhang, Zhen Wang, Jianing Wang, Xiaobing Yu

Automated analysis of temporal changes in multimodal retinal images is critical for the prognostic assessment of ophthalmic diseases. Unlike traditional single-timepoint diagnosis, tracking longitudinal changes across multiple imaging modalities introduces significant data bias challenges: (1) Imbalanced modality samples compromise the integration of knowledge within minority modalities; (2) Heterogeneous visual patterns across modalities undermine the perception of disease-relevant biomarkers. To tackle these issues, we propose a Modality-Incremental Expert Aggregation Network (MoEA-Net), which unifies the inter-modal integration and intra-modal perception for enhanced retinal prognostic prediction. Specifically, we employ the large language model (LLM) with incremental LoRA layers for specific modalities to effectively integrate knowledge from imbalanced data. Besides, we introduce a Spatiotemporal-aware Expert (SAE) module to better perceive both the anatomical structures and longitudinal changes within modalities. By progressively combining the SAE module with incremental LoRA, MoEA-Net supports continual knowledge accumulation and improves accurate reasoning. Experimental results show that MoEA-Net achieves state-of-the-art performance on subretinal fluid change and visual recovery classification tasks, validating its effectiveness.

Hongyang Wang, Yichen Shi, Zhuofu Tao, Yuhao Gao, Liepiao Zhang, Xun Lin, Jun Feng, Xiaochen Yuan, Zitong Yu, Xiaochun Cao

Face anti-spoofing (FAS) is crucial for protecting facial recognition systems from presentation attacks. Previous methods approached this task as a classification problem, lacking interpretability and reasoning behind the predicted results. Recently, multimodal large language models (MLLMs) have shown strong capabilities in perception, reasoning, and decision-making in visual tasks. However, there is currently no universal and comprehensive MLLM and dataset specifically designed for FAS task. To address this gap, we propose FaceShield, a MLLM for FAS, along with the corresponding pre-training and supervised fine-tuning (SFT) datasets, FaceShield-pre10K and FaceShield-sft45K. FaceShield is capable of determining the authenticity of faces, identifying types of spoofing attacks, providing reasoning for its judgments, and detecting attack areas. Specifically, we employ spoof-aware vision perception (SAVP) that incorporates both the original image and auxiliary information based on prior knowledge. We then use an prompt-guided vision token masking (PVTM) strategy to random mask vision tokens, thereby improving the model's generalization ability. We conducted extensive experiments on three benchmark datasets, demonstrating that FaceShield significantly outperforms previous deep learning models and general MLLMs on four FAS tasks, i.e., coarse-grained classification, fine-grained classification, reasoning, and attack localization.

Hongjun Wang, Jiyuan Chen, Xuan Song, Yinqiang Zheng

Post-training quantization offers an efficient pathway to deploy super-resolution models, yet existing methods treat weight and activation quantization independently, missing their critical interplay. Through controlled experiments on SwinIR, we uncover a striking asymmetry: weight quantization primarily degrades structural similarity, while activation quantization disproportionately affects pixel-level accuracy. This stems from their distinct roles—weights encode learned restoration priors for textures and edges, whereas activations carry input-specific intensity information. Building on this insight, we propose HarmoQ, a unified framework that harmonizes quantization across components through three synergistic steps: structural residual calibration proactively adjusts weights to compensate for activation-induced detail loss, harmonized scale optimization analytically balances quantization difficulty via closed-form solutions, and adaptive boundary refinement iteratively maintains this balance during optimization. Experiments show HarmoQ achieves substantial gains under aggressive compression, outperforming prior art by 0.46 dB on Set5 at 2-bit while delivering 3.2× speedup and 4× memory reduction on A100 GPUs. This work provides the first systematic analysis of weight-activation coupling in super-resolution quantization and establishes a principled solution for efficient high-quality image restoration.

Hengkang Wang, Yang Liu, Huidong Liu, Chien-Chih Wang, Yanhui Guo, Hongdong Li, Bryan Wang, Ju Sun

Video restoration (VR) aims to recover high-quality videos from degraded ones. Although recent zero-shot VR methods using pre-trained diffusion models (DMs) show good promise, they suffer from approximation errors during reverse diffusion and insufficient temporal consistency. Moreover, dealing with 3D video data, VR is inherently computationally intensive. In this paper, we advocate viewing the reverse process in DMs as a function and present a novel Maximum a Posterior (MAP) framework that directly parameterizes video frames in the seed space of DMs, eliminating approximation errors. We also introduce strategies to promote bilevel temporal consistency: semantic consistency by leveraging clustering structures in the seed space, and pixel-level consistency by progressive warping with optical flow refinements. Extensive experiments on multiple virtual reality tasks demonstrate superior visual quality and temporal consistency achieved by our method compared to the state-of-the-art.

Hengfei Wang, Zhongqun Zhang, Yihua Cheng, Hyung Jin Chang

Gaze redirection methods aim to generate realistic human face images with controllable eye movement. However, recent methods often struggle with 3D consistency, efficiency, or quality, limiting their practical applications. In this work, we propose RTGaze, a real-time and high-quality gaze redirection method. Our approach learns a gaze-controllable facial representation from face images and gaze prompts, then decodes this representation via neural rendering for gaze redirection. Additionally, we distill face geometric priors from a pretrained 3D portrait generator to enhance generation quality. We evaluate RTGaze both qualitatively and quantitatively, demonstrating state-of-the-art performance in efficiency, redirection accuracy, and image quality across multiple datasets. Our system achieves real-time, 3D-aware gaze redirection with a feedforward network (~0.06 sec/image), making it 800× faster than the previous state-of-the-art 3D-aware methods.

Haoyu Wang, Lei Zhang, Wenrui Liu, Dengyang Jiang, Wei Wei, Chen Ding

Given the inherently costly and time-intensive nature of pixel-level annotation, the generation of synthetic datasets comprising sufficiently diverse synthetic images paired with ground-truth pixel-level annotations has garnered increasing attention recently for training high-performance semantic segmentation models. However, existing methods necessitate to either predict pseudo annotations after image generation or generate images conditioned on manual annotation masks, which incurs image-annotation semantic inconsistency or scalability problem. To migrate both problems with one stone, we present a novel dataset generative diffusion framework for semantic segmentation, termed JoDiffusion. Firstly, given a standard latent diffusion model, JoDiffusion incorporates an independent annotation variational auto-encoder (VAE) network to map annotation masks into the latent space shared by images. Then, the diffusion model is tailored to capture the joint distribution of each image and its annotation mask conditioned on a text prompt. By doing these, JoDiffusion enables simultaneously generating paired images and semantically consistent annotation masks solely conditioned on text prompts, thereby demonstrating superior scalability. Additionally, a mask optimization strategy is developed to mitigate the annotation noise produced during generation. Experiments on Pascal VOC, COCO, and ADE20K datasets show that the annotated dataset generated by JoDiffusion yields substantial performance improvements in semantic segmentation compared to existing methods.

Haotian Wang, Yuzhe Weng, Jun Du, Haoran Xu, Xiaoyan Wu, Shan He, Bing Yin, Cong Liu, Jianqing Gao, Qingfeng Liu

The introduction of diffusion models has brought significant advances to the field of audio-driven talking head generation. However, the extremely slow inference speed severely limits the practical implementation of diffusion-based talking head generation models. In this study, we propose READ, a real-time diffusion-transformer-based talking head generation framework. Our approach first learns a spatiotemporal highly compressed video latent space via a temporal VAE, significantly reducing the token count to accelerate generation. To achieve better audio-visual alignment within this compressed latent space, a pre-trained Speech Autoencoder (SpeechAE) is proposed to generate temporally compressed speech latent codes corresponding to the video latent space. These latent representations are then modeled by a carefully designed Audio-to-Video Diffusion Transformer (A2V-DiT) backbone for efficient talking head synthesis. Furthermore, to ensure temporal consistency and accelerated inference in extended generation, we propose a novel asynchronous noise scheduler (ANS) for both the training and inference processes of our framework. The ANS leverages asynchronous add-noise and asynchronous motion-guided generation in the latent space, ensuring consistency in generated video clips. Experimental results demonstrate that READ outperforms state-of-the-art methods by generating competitive talking head videos with significantly reduced runtime, achieving an optimal balance between quality and speed while maintaining robust metric stability in long-time generation.

Haodong Wang, Tao Zhuo, Xiuwei Zhang, Hanlin Yin, Wencong Wu, Yanning Zhang

Achieving pixel-level registration between SAR and optical images remains a challenging task due to their fundamentally different imaging mechanisms and visual characteristics. Although deep learning has achieved great success in many cross-modal tasks, its performance on SAR-Optical registration tasks is still unsatisfactory. Gradient-based information has traditionally played a crucial role in handcrafted descriptors by highlighting structural differences. However, such gradient cues have not been effectively leveraged in deep learning frameworks for SAR-Optical image matching. To address this gap, we propose SOMA, a dense registration framework that integrates structural gradient priors into deep features and refines alignment through a hybrid matching strategy. Specifically, we introduce the Feature Gradient Enhancer (FGE), which embeds multi-scale, multi-directional gradient filters into the feature space using attention and reconstruction mechanisms to boost feature distinctiveness. Furthermore, we propose the Global-Local Affine-Flow Matcher (GLAM), which combines affine transformation and flow-based refinement within a coarse-to-fine architecture to ensure both structural consistency and local accuracy. Experimental results demonstrate that SOMA significantly improves registration precision, increasing the CMR@1px by 12.29% on the SEN1-2 dataset and 18.50% on the GFGE_SO dataset. In addition, SOMA exhibits strong robustness and generalizes well across diverse scenes and resolutions.

Hao Wang, Eiki Murata, Lingfang Zhang, Ayako Sato, So Fukuda, Ziqi Yin, Wentao Hu, Keisuke Nakao, Yusuke Nakamura, Sebastian Zwirner 等

Recent advances in multimodal large language models (MLLMs) have significantly enhanced video understanding capabilities, opening new possibilities for practical applications. Yet current video benchmarks focus largely on indoor scenes or short-range outdoor activities, leaving the challenges associated with long-distance travel largely unexplored. Mastering extended geospatial-temporal trajectories is critical for next-generation MLLMs, underpinning real-world tasks such as embodied-AI planning and navigation. To bridge this gap, we present VIR-Bench, a novel benchmark consisting of 200 travel videos that frames itinerary reconstruction as a challenging task designed to evaluate and push forward MLLMs' geospatial-temporal intelligence. Experimental results reveal that state-of-the-art MLLMs, including proprietary ones, struggle to achieve high scores, underscoring the difficulty of handling videos that span extended spatial and temporal scales. Moreover, we conduct an in-depth case study in which we develop a prototype travel-planning agent that leverages the insights gained from VIR-Bench. The agent’s markedly improved itinerary recommendations verify that our evaluation protocol not only benchmarks models effectively but also translates into concrete performance gains in user-facing applications.

Hanqing Wang, Shaoyang Wang, Yiming Zhong, Zemin Yang, Jiamin Wang, Zhiqing Cui, Jiahao Yuan, Yifan Han, Mingyu Liu, Yuexin Ma

Affordance grounding focuses on predicting the specific regions of objects that are associated with the actions to be performed by robots. It plays a vital role in the fields of human-robot interaction, human-object interaction, embodied manipulation, and embodied perception. Existing models often neglect the affordance shared among different objects because they lack the Chain-of-Thought(CoT) reasoning abilities, limiting their out-of-domain generalization and explicit reasoning capabilities. To address these challenges, we propose Affordance-R1, the first unified affordance grounding framework that integrates cognitive CoT guided Group Relative Policy Optimization (GRPO) within a reinforcement learning paradigm. Specifically, we designed a sophisticated affordance function, which contains format, perception, and cognition rewards to effectively guide optimization directions. Furthermore, we constructed a high-quality affordance-centric reasoning dataset, ReasonAff, to support training. Trained exclusively via reinforcement learning with GRPO and without explicit reasoning data, Affordance-R1 achieves robust zero-shot generalization and exhibits emergent test-time reasoning capabilities. Comprehensive experiments demonstrate that our model outperforms well-established methods and exhibits open-world generalization.