Xu Zheng

Yunfan Lu, Yiqi Lin, Hao Wu et al.

Image restoration and enhancement is a process of improving the image quality by removing degradations, such as noise, blur, and resolution degradation. Deep learning (DL) has recently been applied to image restoration and enhancement. Due to its ill-posed property, plenty of works have been explored priors to facilitate training deep neural networks (DNNs). However, the importance of priors has not been systematically studied and analyzed by far in the research community. Therefore, this paper serves as the first study that provides a comprehensive overview of recent advancements in priors for deep image restoration and enhancement. Our work covers five primary contents: (1) A theoretical analysis of priors for deep image restoration and enhancement; (2) A hierarchical and structural taxonomy of priors commonly used in the DL-based methods; (3) An insightful discussion on each prior regarding its principle, potential, and applications; (4) A summary of crucial problems by highlighting the potential future directions, especially adopting the large-scale foundation models as prior, to spark more research in the community; (5) An open-source repository that provides a taxonomy of all mentioned works and code links.

Jinjing Zhu, Yunhao Luo, Xu Zheng et al.

In this paper, we strive to answer the question "how to collaboratively learn convolutional neural network (CNN)-based and vision transformer (ViT)-based models by selecting and exchanging the reliable knowledge between them for semantic segmentation?" Accordingly, we propose an online knowledge distillation (KD) framework that can simultaneously learn compact yet effective CNN-based and ViT-based models with two key technical breakthroughs to take full advantage of CNNs and ViT while compensating their limitations. Firstly, we propose heterogeneous feature distillation (HFD) to improve students' consistency in low-layer feature space by mimicking heterogeneous features between CNNs and ViT. Secondly, to facilitate the two students to learn reliable knowledge from each other, we propose bidirectional selective distillation (BSD) that can dynamically transfer selective knowledge. This is achieved by 1) region-wise BSD determining the directions of knowledge transferred between the corresponding regions in the feature space and 2) pixel-wise BSD discerning which of the prediction knowledge to be transferred in the logit space. Extensive experiments on three benchmark datasets demonstrate that our proposed framework outperforms the state-of-the-art online distillation methods by a large margin, and shows its efficacy in learning collaboratively between ViT-based and CNN-based models.

Xu Zheng, Tianbo Pan, Yunhao Luo et al.

Endeavors have been recently made to transfer knowledge from the labeled pinhole image domain to the unlabeled panoramic image domain via Unsupervised Domain Adaptation (UDA). The aim is to tackle the domain gaps caused by the style disparities and distortion problem from the non-uniformly distributed pixels of equirectangular projection (ERP). Previous works typically focus on transferring knowledge based on geometric priors with specially designed multi-branch network architectures. As a result, considerable computational costs are induced, and meanwhile, their generalization abilities are profoundly hindered by the variation of distortion among pixels. In this paper, we find that the pixels' neighborhood regions of the ERP indeed introduce less distortion. Intuitively, we propose a novel UDA framework that can effectively address the distortion problems for panoramic semantic segmentation. In comparison, our method is simpler, easier to implement, and more computationally efficient. Specifically, we propose distortion-aware attention (DA) capturing the neighboring pixel distribution without using any geometric constraints. Moreover, we propose a class-wise feature aggregation (CFA) module to iteratively update the feature representations with a memory bank. As such, the feature similarity between two domains can be consistently optimized. Extensive experiments show that our method achieves new state-of-the-art performance while remarkably reducing 80% parameters.

Xu Zheng, Yexin Liu, Yunfan Lu et al.

Event cameras are bio-inspired sensors that capture the per-pixel intensity changes asynchronously and produce event streams encoding the time, pixel position, and polarity (sign) of the intensity changes. Event cameras possess a myriad of advantages over canonical frame-based cameras, such as high temporal resolution, high dynamic range, low latency, etc. Being capable of capturing information in challenging visual conditions, event cameras have the potential to overcome the limitations of frame-based cameras in the computer vision and robotics community. In very recent years, deep learning (DL) has been brought to this emerging field and inspired active research endeavors in mining its potential. However, there is still a lack of taxonomies in DL techniques for event-based vision. We first scrutinize the typical event representations with quality enhancement methods as they play a pivotal role as inputs to the DL models. We then provide a comprehensive survey of existing DL-based methods by structurally grouping them into two major categories: 1) image/video reconstruction and restoration; 2) event-based scene understanding and 3D vision. We conduct benchmark experiments for the existing methods in some representative research directions, i.e., image reconstruction, deblurring, and object recognition, to identify some critical insights and problems. Finally, we have discussions regarding the challenges and provide new perspectives for inspiring more research studies.

Xu Zheng, Farhad Shirani, Tianchun Wang et al.

Graph Neural Networks (GNNs) are neural models that leverage the dependency structure in graphical data via message passing among the graph nodes. GNNs have emerged as pivotal architectures in analyzing graph-structured data, and their expansive application in sensitive domains requires a comprehensive understanding of their decision-making processes -- necessitating a framework for GNN explainability. An explanation function for GNNs takes a pre-trained GNN along with a graph as input, to produce a `sufficient statistic' subgraph with respect to the graph label. A main challenge in studying GNN explainability is to provide fidelity measures that evaluate the performance of these explanation functions. This paper studies this foundational challenge, spotlighting the inherent limitations of prevailing fidelity metrics, including $Fid_+$, $Fid_-$, and $Fid_Δ$. Specifically, a formal, information-theoretic definition of explainability is introduced and it is shown that existing metrics often fail to align with this definition across various statistical scenarios. The reason is due to potential distribution shifts when subgraphs are removed in computing these fidelity measures. Subsequently, a robust class of fidelity measures are introduced, and it is shown analytically that they are resilient to distribution shift issues and are applicable in a wide range of scenarios. Extensive empirical analysis on both synthetic and real datasets are provided to illustrate that the proposed metrics are more coherent with gold standard metrics. The source code is available at https://trustai4s-lab.github.io/fidelity.

Xu Zheng, Yunhao Luo, Chong Fu et al.

The popular methods for semi-supervised semantic segmentation mostly adopt a unitary network model using convolutional neural networks (CNNs) and enforce consistency of the model's predictions over perturbations applied to the inputs or model. However, such a learning paradigm suffers from two critical limitations: a) learning the discriminative features for the unlabeled data; b) learning both global and local information from the whole image. In this paper, we propose a novel Semi-supervised Learning (SSL) approach, called Transformer-CNN Cohort (TCC), that consists of two students with one based on the vision transformer (ViT) and the other based on the CNN. Our method subtly incorporates the multi-level consistency regularization on the predictions and the heterogeneous feature spaces via pseudo-labeling for the unlabeled data. First, as the inputs of the ViT student are image patches, the feature maps extracted encode crucial class-wise statistics. To this end, we propose class-aware feature consistency distillation (CFCD) that first leverages the outputs of each student as the pseudo labels and generates class-aware feature (CF) maps for knowledge transfer between the two students. Second, as the ViT student has more uniform representations for all layers, we propose consistency-aware cross distillation (CCD) to transfer knowledge between the pixel-wise predictions from the cohort. We validate the TCC framework on Cityscapes and Pascal VOC 2012 datasets, which outperforms existing SSL methods by a large margin.

Hongfei Zhang, Harold Haodong Chen, Chenfei Liao et al.

Existing video depth estimation faces a fundamental trade-off: generative models suffer from stochastic geometric hallucinations and scale drift, while discriminative models demand massive labeled datasets to resolve semantic ambiguities. To break this impasse, we present DVD, the first framework to deterministically adapt pre-trained video diffusion models into single-pass depth regressors. Specifically, DVD features three core designs: (i) repurposing the diffusion timestep as a structural anchor to balance global stability with high-frequency details; (ii) latent manifold rectification (LMR) to mitigate regression-induced over-smoothing, enforcing differential constraints to restore sharp boundaries and coherent motion; and (iii) global affine coherence, an inherent property bounding inter-window divergence, which enables seamless long-video inference without requiring complex temporal alignment. Extensive experiments demonstrate that DVD achieves state-of-the-art zero-shot performance across benchmarks. Furthermore, DVD successfully unlocks the profound geometric priors implicit in video foundation models using 163x less task-specific data than leading baselines. Notably, we fully release our pipeline, providing the whole training suite for SOTA video depth estimation to benefit the open-source community.

Yuanfan Zheng, Kunyu Peng, Xu Zheng et al.

Cross-domain panoramic semantic segmentation has attracted growing interest as it enables comprehensive 360° scene understanding for real-world applications. However, it remains particularly challenging due to severe geometric Field of View (FoV) distortions and inconsistent open-set semantics across domains. In this work, we formulate an open-set domain adaptation setting, and propose Extrapolative Domain Adaptive Panoramic Segmentation (EDA-PSeg) framework that trains on local perspective views and tests on full 360° panoramic images, explicitly tackling both geometric FoV shifts across domains and semantic uncertainty arising from previously unseen classes. To this end, we propose the Euler-Margin Attention (EMA), which introduces an angular margin to enhance viewpoint-invariant semantic representation, while performing amplitude and phase modulation to improve generalization toward unseen classes. Additionally, we design the Graph Matching Adapter (GMA), which builds high-order graph relations to align shared semantics across FoV shifts while effectively separating novel categories through structural adaptation. Extensive experiments on four benchmark datasets under camera-shift, weather-condition, and open-set scenarios demonstrate that EDA-PSeg achieves state-of-the-art performance, robust generalization to diverse viewing geometries, and resilience under varying environmental conditions. The code is available at https://github.com/zyfone/EDA-PSeg.

Ye Pan, Chi Kit Wong, Yuanhuiyi Lyu et al.

Multimodal Large Language Models (MLLMs) have demonstrated remarkable video reasoning capabilities across diverse tasks. However, their ability to understand human intent at a fine-grained level in egocentric videos remains largely unexplored. Existing benchmarks focus primarily on episode-level intent reasoning, overlooking the finer granularity of step-level intent understanding. Yet applications such as intelligent assistants, robotic imitation learning, and augmented reality guidance require understanding not only what a person is doing at each step, but also why and what comes next, in order to provide timely and context-aware support. To this end, we introduce EgoIntent, a step-level intent understanding benchmark for egocentric videos. It comprises 3,014 steps spanning 15 diverse indoor and outdoor daily-life scenarios, and evaluates models on three complementary dimensions: local intent (What), global intent (Why), and next-step plan (Next). Crucially, each clip is truncated immediately before the key outcome of the queried step (e.g., contact or grasp) occurs and contains no frames from subsequent steps, preventing future-frame leakage and enabling a clean evaluation of anticipatory step understanding and next-step planning. We evaluate 15 MLLMs, including both state-of-the-art closed-source and open-source models. Even the best-performing model achieves an average score of only 33.31 across the three intent dimensions, underscoring that step-level intent understanding in egocentric videos remains a highly challenging problem that calls for further investigation.

Xu Zheng, Jinjing Zhu, Yexin Liu et al.

The ability of scene understanding has sparked active research for panoramic image semantic segmentation. However, the performance is hampered by distortion of the equirectangular projection (ERP) and a lack of pixel-wise annotations. For this reason, some works treat the ERP and pinhole images equally and transfer knowledge from the pinhole to ERP images via unsupervised domain adaptation (UDA). However, they fail to handle the domain gaps caused by: 1) the inherent differences between camera sensors and captured scenes; 2) the distinct image formats (e.g., ERP and pinhole images). In this paper, we propose a novel yet flexible dual-path UDA framework, DPPASS, taking ERP and tangent projection (TP) images as inputs. To reduce the domain gaps, we propose cross-projection and intra-projection training. The cross-projection training includes tangent-wise feature contrastive training and prediction consistency training. That is, the former formulates the features with the same projection locations as positive examples and vice versa, for the models' awareness of distortion, while the latter ensures the consistency of cross-model predictions between the ERP and TP. Moreover, adversarial intra-projection training is proposed to reduce the inherent gap, between the features of the pinhole images and those of the ERP and TP images, respectively. Importantly, the TP path can be freely removed after training, leading to no additional inference cost. Extensive experiments on two benchmarks show that our DPPASS achieves +1.06$\%$ mIoU increment than the state-of-the-art approaches.

Xu Zheng, Yunhao Luo, Pengyuan Zhou et al.

In this paper, we tackle a new problem: how to transfer knowledge from the pre-trained cumbersome yet well-performed CNN-based model to learn a compact Vision Transformer (ViT)-based model while maintaining its learning capacity? Due to the completely different characteristics of ViT and CNN and the long-existing capacity gap between teacher and student models in Knowledge Distillation (KD), directly transferring the cross-model knowledge is non-trivial. To this end, we subtly leverage the visual and linguistic-compatible feature character of ViT (i.e., student), and its capacity gap with the CNN (i.e., teacher) and propose a novel CNN-to-ViT KD framework, dubbed C2VKD. Importantly, as the teacher's features are heterogeneous to those of the student, we first propose a novel visual-linguistic feature distillation (VLFD) module that explores efficient KD among the aligned visual and linguistic-compatible representations. Moreover, due to the large capacity gap between the teacher and student and the inevitable prediction errors of the teacher, we then propose a pixel-wise decoupled distillation (PDD) module to supervise the student under the combination of labels and teacher's predictions from the decoupled target and non-target classes. Experiments on three semantic segmentation benchmark datasets consistently show that the increment of mIoU of our method is over 200% of the SoTA KD methods

Jiahang Cao, Xu Zheng, Yuanhuiyi Lyu et al.

The ability to detect objects in all lighting (i.e., normal-, over-, and under-exposed) conditions is crucial for real-world applications, such as self-driving.Traditional RGB-based detectors often fail under such varying lighting conditions.Therefore, recent works utilize novel event cameras to supplement or guide the RGB modality; however, these methods typically adopt asymmetric network structures that rely predominantly on the RGB modality, resulting in limited robustness for all-day detection. In this paper, we propose EOLO, a novel object detection framework that achieves robust and efficient all-day detection by fusing both RGB and event modalities. Our EOLO framework is built based on a lightweight spiking neural network (SNN) to efficiently leverage the asynchronous property of events. Buttressed by it, we first introduce an Event Temporal Attention (ETA) module to learn the high temporal information from events while preserving crucial edge information. Secondly, as different modalities exhibit varying levels of importance under diverse lighting conditions, we propose a novel Symmetric RGB-Event Fusion (SREF) module to effectively fuse RGB-Event features without relying on a specific modality, thus ensuring a balanced and adaptive fusion for all-day detection. In addition, to compensate for the lack of paired RGB-Event datasets for all-day training and evaluation, we propose an event synthesis approach based on the randomized optical flow that allows for directly generating the event frame from a single exposure image. We further build two new datasets, E-MSCOCO and E-VOC based on the popular benchmarks MSCOCO and PASCAL VOC. Extensive experiments demonstrate that our EOLO outperforms the state-of-the-art detectors,e.g.,RENet,by a substantial margin (+3.74% mAP50) in all lighting conditions.Our code and datasets will be available at https://vlislab22.github.io/EOLO/

Xu Zheng, Yuanhuiyi Lyu, Lin Wang

Image modality is not perfect as it often fails in certain conditions, e.g., night and fast motion. This significantly limits the robustness and versatility of existing multi-modal (i.e., Image+X) semantic segmentation methods when confronting modality absence or failure, as often occurred in real-world applications. Inspired by the open-world learning capability of multi-modal vision-language models (MVLMs), we explore a new direction in learning the modality-agnostic representation via knowledge distillation (KD) from MVLMs. Intuitively, we propose Any2Seg, a novel framework that can achieve robust segmentation from any combination of modalities in any visual conditions. Specifically, we first introduce a novel language-guided semantic correlation distillation (LSCD) module to transfer both inter-modal and intra-modal semantic knowledge in the embedding space from MVLMs, e.g., LanguageBind. This enables us to minimize the modality gap and alleviate semantic ambiguity to combine any modalities in any visual conditions. Then, we introduce a modality-agnostic feature fusion (MFF) module that reweights the multi-modal features based on the inter-modal correlation and selects the fine-grained feature. This way, our Any2Seg finally yields an optimal modality-agnostic representation. Extensive experiments on two benchmarks with four modalities demonstrate that Any2Seg achieves the state-of-the-art under the multi-modal setting (+3.54 mIoU) and excels in the challenging modality-incomplete setting(+19.79 mIoU).

Jiazhou Zhou, Xu Zheng, Yuanhuiyi Lyu et al.

In this paper, we propose EventBind, a novel and effective framework that unleashes the potential of vision-language models (VLMs) for event-based recognition to compensate for the lack of large-scale event-based datasets. In particular, due to the distinct modality gap with the image-text data and the lack of large-scale datasets, learning a common representation space for images, texts, and events is non-trivial.Intuitively, we need to address two key challenges: 1) how to generalize CLIP's visual encoder to event data while fully leveraging events' unique properties, e.g., sparsity and high temporal resolution; 2) how to effectively align the multi-modal embeddings, i.e., image, text, and events. Accordingly, we first introduce a novel event encoder that subtly models the temporal information from events and meanwhile, generates event prompts for modality bridging. We then design a text encoder that generates content prompts and utilizes hybrid text prompts to enhance EventBind's generalization ability across diverse datasets.With the proposed event encoder, text encoder, and image encoder, a novel Hierarchical Triple Contrastive Alignment (HTCA) module is introduced to jointly optimize the correlation and enable efficient knowledge transfer among the three modalities. We evaluate various settings, including fine-tuning and few-shot on three benchmarks, and our EventBind achieves new state-of-the-art accuracy compared with the previous methods, such as on N-Caltech101 (+5.34% and +1.70%) and N-Imagenet (+5.65% and +1.99%) with fine-tuning and 20-shot settings, respectively. Moreover, our EventBind can be flexibly extended to the event retrieval task using text or image queries, showing plausible performance. Project page:https://vlislab22.github.io/EventBind/.

Chih-Ting Liao, Xi Xiao, Chunlei Meng et al.

Multimodal large language models (MLLMs) have advanced static visual--spatial reasoning, yet they often fail to preserve long-horizon spatial coherence in embodied settings where beliefs must be continuously revised from egocentric observations under environmental change. We introduce SpaMEM (Spatial Memory from Action Sequences), a large-scale diagnostic benchmark that isolates the mechanics of spatial belief evolution via action-conditioned scene transformations (spawn, place, remove) over long interaction horizons. SpaMEM is built on a physically grounded dataset with 10,601,392 high-fidelity images across four modalities (RGB, depth, instance, semantic segmentation), collected from 25,000+ interaction sequences in 1,000 procedurally generated houses. We formalize embodied spatial reasoning as a three-level hierarchy with 15 diagnostic tasks: Level 1 measures atomic spatial perception from single observations; Level 2 probes temporal reasoning with oracle textual state histories to factor out perceptual noise; and Level 3 requires end-to-end belief maintenance from raw visual streams under the same task dimensions. We further evaluate both short-term (step-wise) updates and long-term (episodic) reconstruction. Benchmarking representative open-source VLM families reveals a consistent stacked bottleneck: coordinate-consistent grounding remains a hard ceiling, and the sharp collapse from Level 2 to Level 3 exposes a pronounced symbolic scaffolding dependency, where models succeed with text-based bookkeeping but struggle to sustain robust visual memory. SpaMEM provides a granular diagnostic standard and motivates explicit mechanisms for state representation, belief revision, and long-horizon episodic integration.

Xu Zheng, Yuanhuiyi Lyu, Jiazhou Zhou et al.

Fusing an arbitrary number of modalities is vital for achieving robust multi-modal fusion of semantic segmentation yet remains less explored to date. Recent endeavors regard RGB modality as the center and the others as the auxiliary, yielding an asymmetric architecture with two branches. However, the RGB modality may struggle in certain circumstances, e.g., nighttime, while others, e.g., event data, own their merits; thus, it is imperative for the fusion model to discern robust and fragile modalities, and incorporate the most robust and fragile ones to learn a resilient multi-modal framework. To this end, we propose a novel method, named MAGIC, that can be flexibly paired with various backbones, ranging from compact to high-performance models. Our method comprises two key plug-and-play modules. Firstly, we introduce a multi-modal aggregation module to efficiently process features from multi-modal batches and extract complementary scene information. On top, a unified arbitrary-modal selection module is proposed to utilize the aggregated features as the benchmark to rank the multi-modal features based on the similarity scores. This way, our method can eliminate the dependence on RGB modality and better overcome sensor failures while ensuring the segmentation performance. Under the commonly considered multi-modal setting, our method achieves state-of-the-art performance while reducing the model parameters by 60%. Moreover, our method is superior in the novel modality-agnostic setting, where it outperforms prior arts by a large margin of +19.41% mIoU

Jialei Chen, Daisuke Deguchi, Chenkai Zhang et al.

Generalized Zero-shot Semantic Segmentation aims to segment both seen and unseen categories only under the supervision of the seen ones. To tackle this, existing methods adopt the large-scale Vision Language Models (VLMs) which obtain outstanding zero-shot performance. However, as the VLMs are designed for classification tasks, directly adapting the VLMs may lead to sub-optimal performance. Consequently, we propose CLIP-ZSS (Zero-shot Semantic Segmentation), a simple but effective training framework that enables any image encoder designed for closed-set segmentation applied in zero-shot and open-vocabulary tasks in testing without combining with VLMs or inserting new modules. CLIP-ZSS consists of two key modules: Global Learning Module (GLM) and Pixel Learning Module (PLM). GLM is proposed to probe the knowledge from the CLIP visual encoder by pulling the CLS token and the dense features from the image encoder of the same image and pushing others apart. Moreover, to enhance the ability to discriminate unseen categories, PLM consisting of pseudo labels and weight generation is designed. To generate semantically discriminated pseudo labels, a multi-scale K-Means with mask fusion working on the dense tokens is proposed. In pseudo weight generation, a synthesizer generating pseudo semantic features for the unannotated area is introduced. Experiments on three benchmarks show large performance gains compared with SOTA methods.

Weiming Zhang, Yexin Liu, Xu Zheng et al.

This paper presents GoodSAM++, a novel framework utilizing the powerful zero-shot instance segmentation capability of SAM (i.e., teacher) to learn a compact panoramic semantic segmentation model, i.e., student, without requiring any labeled data. GoodSAM++ addresses two critical challenges: 1) SAM's inability to provide semantic labels and inherent distortion problems of panoramic images; 2) the significant capacity disparity between SAM and the student. The `out-of-the-box' insight of GoodSAM++ is to introduce a teacher assistant (TA) to provide semantic information for SAM, integrated with SAM to obtain reliable pseudo semantic maps to bridge both domain and capacity gaps. To make this possible, we first propose a Distortion-Aware Rectification (DARv2) module to address the domain gap. It effectively mitigates the object deformation and distortion problem in panoramic images to obtain pseudo semantic maps. We then introduce a Multi-level Knowledge Adaptation (MKA) module to efficiently transfer the semantic information from the TA and pseudo semantic maps to our compact student model, addressing the significant capacity gap. We conduct extensive experiments on both outdoor and indoor benchmark datasets, showing that our GoodSAM++ achieves a remarkable performance improvement over the state-of-the-art (SOTA) domain adaptation methods. Moreover, diverse open-world scenarios demonstrate the generalization capacity of our GoodSAM++. Last but not least, our most lightweight student model achieves comparable performance to the SOTA models with only 3.7 million parameters.

Zibo Zhao, Zeqiang Lai, Qingxiang Lin et al.

We present Hunyuan3D 2.0, an advanced large-scale 3D synthesis system for generating high-resolution textured 3D assets. This system includes two foundation components: a large-scale shape generation model -- Hunyuan3D-DiT, and a large-scale texture synthesis model -- Hunyuan3D-Paint. The shape generative model, built on a scalable flow-based diffusion transformer, aims to create geometry that properly aligns with a given condition image, laying a solid foundation for downstream applications. The texture synthesis model, benefiting from strong geometric and diffusion priors, produces high-resolution and vibrant texture maps for either generated or hand-crafted meshes. Furthermore, we build Hunyuan3D-Studio -- a versatile, user-friendly production platform that simplifies the re-creation process of 3D assets. It allows both professional and amateur users to manipulate or even animate their meshes efficiently. We systematically evaluate our models, showing that Hunyuan3D 2.0 outperforms previous state-of-the-art models, including the open-source models and closed-source models in geometry details, condition alignment, texture quality, and etc. Hunyuan3D 2.0 is publicly released in order to fill the gaps in the open-source 3D community for large-scale foundation generative models. The code and pre-trained weights of our models are available at: https://github.com/Tencent/Hunyuan3D-2

Xu Zheng, Ling Wang, Kanghao Chen et al.

Recently, electroencephalography (EEG) signals have been actively incorporated to decode brain activity to visual or textual stimuli and achieve object recognition in multi-modal AI. Accordingly, endeavors have been focused on building EEG-based datasets from visual or textual single-modal stimuli. However, these datasets offer limited EEG epochs per category, and the complex semantics of stimuli presented to participants compromise their quality and fidelity in capturing precise brain activity. The study in neuroscience unveils that the relationship between visual and textual stimulus in EEG recordings provides valuable insights into the brain's ability to process and integrate multi-modal information simultaneously. Inspired by this, we propose a novel large-scale multi-modal dataset, named EIT-1M, with over 1 million EEG-image-text pairs. Our dataset is superior in its capacity of reflecting brain activities in simultaneously processing multi-modal information. To achieve this, we collected data pairs while participants viewed alternating sequences of visual-textual stimuli from 60K natural images and category-specific texts. Common semantic categories are also included to elicit better reactions from participants' brains. Meanwhile, response-based stimulus timing and repetition across blocks and sessions are included to ensure data diversity. To verify the effectiveness of EIT-1M, we provide an in-depth analysis of EEG data captured from multi-modal stimuli across different categories and participants, along with data quality scores for transparency. We demonstrate its validity on two tasks: 1) EEG recognition from visual or textual stimuli or both and 2) EEG-to-visual generation.

Xu Zheng, Lin Wang

In this paper, we address the challenging problem of cross-modal (image-to-events) adaptation for event-based recognition without accessing any labeled source image data. This task is arduous due to the substantial modality gap between images and events. With only a pre-trained source model available, the key challenge lies in extracting knowledge from this model and effectively transferring knowledge to the event-based domain. Inspired by the natural ability of language to convey semantics across different modalities, we propose EventDance++, a novel framework that tackles this unsupervised source-free cross-modal adaptation problem from a language-guided perspective. We introduce a language-guided reconstruction-based modality bridging (L-RMB) module, which reconstructs intensity frames from events in a self-supervised manner. Importantly, it leverages a vision-language model to provide further supervision, enriching the surrogate images and enhancing modality bridging. This enables the creation of surrogate images to extract knowledge (i.e., labels) from the source model. On top, we propose a multi-representation knowledge adaptation (MKA) module to transfer knowledge to target models, utilizing multiple event representations to capture the spatiotemporal characteristics of events fully. The L-RMB and MKA modules are jointly optimized to achieve optimal performance in bridging the modality gap. Experiments on three benchmark datasets demonstrate that EventDance++ performs on par with methods that utilize source data, validating the effectiveness of our language-guided approach in event-based recognition.

Yibo Yan, Jiahao Huo, Guanbo Feng et al.

With the rapid proliferation of multimodal information, Visual Document Retrieval (VDR) has emerged as a critical frontier in bridging the gap between unstructured visually rich data and precise information acquisition. Unlike traditional natural image retrieval, visual documents exhibit unique characteristics defined by dense textual content, intricate layouts, and fine-grained semantic dependencies. This paper presents the first comprehensive survey of the VDR landscape, specifically through the lens of the Multimodal Large Language Model (MLLM) era. We begin by examining the benchmark landscape, and subsequently dive into the methodological evolution, categorizing approaches into three primary aspects: multimodal embedding models, multimodal reranker models, and the integration of Retrieval-Augmented Generation (RAG) and Agentic systems for complex document intelligence. Finally, we identify persistent challenges and outline promising future directions, aiming to provide a clear roadmap for future multimodal document intelligence.

Yuanhuiyi Lyu, Kaiyu Lei, Ziqiao Weng et al.

Reasoning-based text-to-image (T2I) generation requires models to interpret complex prompts accurately. Existing reasoning frameworks can be broadly categorized into two types: (1) Text-Only Reasoning, which is computationally efficient but lacks access to visual context, often resulting in the omission of critical spatial and visual elements; and (2) Text-Image Interleaved Reasoning, which leverages a T2I generator to provide visual references during the reasoning process. While this approach enhances visual grounding, it incurs substantial computational costs and constrains the reasoning capacity of MLLMs to the representational limitations of the generator. To this end, we propose StruVis, a novel framework that enhances T2I generation through Thinking with Structured Vision. Instead of relying on intermediate image generation, StruVis employs text-based structured visual representations as intermediate reasoning states, thereby enabling the MLLM to effectively "perceive" visual structure within a purely text-based reasoning process. Powered by this, the reasoning potential for T2I generation of the MLLM is unlocked through structured-vision-guided reasoning. Additionally, as a generator-agnostic reasoning framework, our proposed StruVis can be seamlessly integrated with diverse T2I generators and efficiently enhance their performance in reasoning-based T2I generation. Extensive experiments demonstrate that StruVis achieves significant performance improvements on reasoning-based T2I benchmarks, e.g., a 4.61% gain on T2I-ReasonBench and a 4% gain on WISE.

Lutao Jiang, Xu Zheng, Yuanhuiyi Lyu et al.

Text-to-3D synthesis has recently seen intriguing advances by combining the text-to-image priors with 3D representation methods, e.g., 3D Gaussian Splatting (3D GS), via Score Distillation Sampling (SDS). However, a hurdle of existing methods is the low efficiency, per-prompt optimization for a single 3D object. Therefore, it is imperative for a paradigm shift from per-prompt optimization to feed-forward generation for any unseen text prompts, which yet remains challenging. An obstacle is how to directly generate a set of millions of 3D Gaussians to represent a 3D object. This paper presents BrightDreamer, an end-to-end feed-forward approach that can achieve generalizable and fast (77 ms) text-to-3D generation. Our key idea is to formulate the generation process as estimating the 3D deformation from an anchor shape with predefined positions. For this, we first propose a Text-guided Shape Deformation (TSD) network to predict the deformed shape and its new positions, used as the centers (one attribute) of 3D Gaussians. To estimate the other four attributes (i.e., scaling, rotation, opacity, and SH), we then design a novel Text-guided Triplane Generator (TTG) to generate a triplane representation for a 3D object. The center of each Gaussian enables us to transform the spatial feature into the four attributes. The generated 3D Gaussians can be finally rendered at 705 frames per second. Extensive experiments demonstrate the superiority of our method over existing methods. Also, BrightDreamer possesses a strong semantic understanding capability even for complex text prompts. The code is available in the project page.

Zichuan Liu, Tianchun Wang, Jimeng Shi et al.

Explaining deep learning models operating on time series data is crucial in various applications of interest which require interpretable and transparent insights from time series signals. In this work, we investigate this problem from an information theoretic perspective and show that most existing measures of explainability may suffer from trivial solutions and distributional shift issues. To address these issues, we introduce a simple yet practical objective function for time series explainable learning. The design of the objective function builds upon the principle of information bottleneck (IB), and modifies the IB objective function to avoid trivial solutions and distributional shift issues. We further present TimeX++, a novel explanation framework that leverages a parametric network to produce explanation-embedded instances that are both in-distributed and label-preserving. We evaluate TimeX++ on both synthetic and real-world datasets comparing its performance against leading baselines, and validate its practical efficacy through case studies in a real-world environmental application. Quantitative and qualitative evaluations show that TimeX++ outperforms baselines across all datasets, demonstrating a substantial improvement in explanation quality for time series data. The source code is available at \url{https://github.com/zichuan-liu/TimeXplusplus}.

Jiahao Huo, Yibo Yan, Xu Zheng et al.

Recent progress in Machine Unlearning (MU) has introduced solutions for the selective removal of private or sensitive information encoded within deep neural networks. Nonetheless, MU for Multimodal Large Language Models (MLLMs) remains in its nascent phase. Therefore, we propose to reformulate the task of multimodal MU in the era of MLLMs, which aims to erase only the visual patterns associated with a given entity while preserving the corresponding textual knowledge encoded within the original parameters of the language model backbone. Furthermore, we develop a novel geometry-constrained gradient ascent method MMUnlearner. It updates the weights of MLLMs with a weight saliency map jointly restricted by the remaining concepts and textual knowledge during unlearning, thereby preserving parameters essential for non-target knowledge. Extensive experiments demonstrate that MMUnlearner surpasses baselines that finetuning MLLMs with VQA data directly through Gradient Ascent (GA) or Negative Preference Optimization (NPO), across all evaluation dimensions. Our code can be found in [this URL](https://github.com/Z1zs/MMUnlearner).

Chenfei Liao, Kaiyu Lei, Xu Zheng et al.

Multi-modal semantic segmentation (MMSS) addresses the limitations of single-modality data by integrating complementary information across modalities. Despite notable progress, a significant gap persists between research and real-world deployment due to variability and uncertainty in multi-modal data quality. Robustness has thus become essential for practical MMSS applications. However, the absence of standardized benchmarks for evaluating robustness hinders further advancement. To address this, we first survey existing MMSS literature and categorize representative methods to provide a structured overview. We then introduce a robustness benchmark that evaluates MMSS models under three scenarios: Entire-Missing Modality (EMM), Random-Missing Modality (RMM), and Noisy Modality (NM). From a probabilistic standpoint, we model modality failure under two conditions: (1) all damaged combinations are equally probable; (2) each modality fails independently following a Bernoulli distribution. Based on these, we propose four metrics-$mIoU^{Avg}_{EMM}$, $mIoU^{E}_{EMM}$, $mIoU^{Avg}_{RMM}$, and $mIoU^{E}_{RMM}$-to assess model robustness under EMM and RMM. This work provides the first dedicated benchmark for MMSS robustness, offering new insights and tools to advance the field. Source code is available at https://github.com/Chenfei-Liao/Multi-Modal-Semantic-Segmentation-Robustness-Benchmark.

Jiayi Zhao, Fei Teng, Kai Luo et al.

The perception capability of robotic systems relies on the richness of the dataset. Although Segment Anything Model 2 (SAM2), trained on large datasets, demonstrates strong perception potential in perception tasks, its inherent training paradigm prevents it from being suitable for RGB-T tasks. To address these challenges, we propose SHIFNet, a novel SAM2-driven Hybrid Interaction Paradigm that unlocks the potential of SAM2 with linguistic guidance for efficient RGB-Thermal perception. Our framework consists of two key components: (1) Semantic-Aware Cross-modal Fusion (SACF) module that dynamically balances modality contributions through text-guided affinity learning, overcoming SAM2's inherent RGB bias; (2) Heterogeneous Prompting Decoder (HPD) that enhances global semantic information through a semantic enhancement module and then combined with category embeddings to amplify cross-modal semantic consistency. With 32.27M trainable parameters, SHIFNet achieves state-of-the-art segmentation performance on public benchmarks, reaching 89.8% on PST900 and 67.8% on FMB, respectively. The framework facilitates the adaptation of pre-trained large models to RGB-T segmentation tasks, effectively mitigating the high costs associated with data collection while endowing robotic systems with comprehensive perception capabilities. The source code will be made publicly available at https://github.com/iAsakiT3T/SHIFNet.

Zihao Dongfang, Xu Zheng, Ziqiao Weng et al.

The 180x360 omnidirectional field of view captured by 360-degree cameras enables their use in a wide range of applications such as embodied AI and virtual reality. Although recent advances in multimodal large language models (MLLMs) have shown promise in visual-spatial reasoning, most studies focus on standard pinhole-view images, leaving omnidirectional perception largely unexplored. In this paper, we ask: Are MLLMs ready for omnidirectional spatial reasoning? To investigate this, we introduce OSR-Bench, the first benchmark specifically designed for this setting. OSR-Bench includes over 153,000 diverse question-answer pairs grounded in high-fidelity panoramic indoor scene maps. It covers key reasoning types including object counting, relative distance, and direction. We also propose a negative sampling strategy that inserts non-existent objects into prompts to evaluate hallucination and grounding robustness. For fine-grained analysis, we design a two-stage evaluation framework assessing both cognitive map generation and QA accuracy using rotation-invariant matching and a combination of rule-based and LLM-based metrics. We evaluate eight state-of-the-art MLLMs, including GPT-4o, Gemini 1.5 Pro, and leading open-source models under zero-shot settings. Results show that current models struggle with spatial reasoning in panoramic contexts, highlighting the need for more perceptually grounded MLLMs. OSR-Bench and code will be released at: https://huggingface.co/datasets/UUUserna/OSR-Bench

Dingcheng Zhen, Xu Zheng, Ruixin Zhang et al.

Autoregressive (AR) diffusion models offer a promising framework for sequential generation tasks such as video synthesis by combining diffusion modeling with causal inference. Although they support streaming generation, existing AR diffusion methods struggle to scale efficiently. In this paper, we identify two key challenges in hour-scale real-time human animation. First, most forcing strategies propagate sample-level representations with mismatched diffusion states, causing inconsistent learning signals and unstable convergence. Second, historical representations grow unbounded and lack structure, preventing effective reuse of cached states and severely limiting inference efficiency. To address these challenges, we propose Neighbor Forcing, a diffusion-step-consistent AR formulation that propagates temporally adjacent frames as latent neighbors under the same noise condition. This design provides a distribution-aligned and stable learning signal while preserving drifting throughout the AR chain. Building upon this, we introduce a structured ConvKV memory mechanism that compresses the keys and values in causal attention into a fixed-length representation, enabling constant-memory inference and truly infinite video generation without relying on short-term motion-frame memory. Extensive experiments demonstrate that our approach significantly improves training convergence, hour-scale generation quality, and inference efficiency compared to existing AR diffusion methods. Numerically, LiveAct enables hour-scale real-time human animation and supports 20 FPS real-time streaming inference on as few as two NVIDIA H100 or H200 GPUs. Quantitative results demonstrate that our method attains state-of-the-art performance in lip-sync accuracy, human animation quality, and emotional expressiveness, with the lowest inference cost.

Emily Muller, Xu Zheng, Jer Hayes

Generative models have been found effective for data synthesis due to their ability to capture complex underlying data distributions. The quality of generated data from these models is commonly evaluated by visual inspection for image datasets or downstream analytical tasks for tabular datasets. These evaluation methods neither measure the implicit data distribution nor consider the data privacy issues, and it remains an open question of how to compare and rank different generative models. Medical data can be sensitive, so it is of great importance to draw privacy concerns of patients while maintaining the data utility of the synthetic dataset. Beyond the utility evaluation, this work outlines two metrics called Similarity and Uniqueness for sample-wise assessment of synthetic datasets. We demonstrate the proposed notions with several state-of-the-art generative models to synthesise Cystic Fibrosis (CF) patients' electronic health records (EHRs), observing that the proposed metrics are suitable for synthetic data evaluation and generative model comparison.

Xiang Gu, Liming Lu, Xu Zheng et al.

Adversarial attacks pose a significant threat to learning-based 3D point cloud models, critically undermining their reliability in security-sensitive applications. Existing defense methods often suffer from (1) high computational overhead and (2) poor generalization ability across diverse attack types. To bridge these gaps, we propose a novel yet efficient teacher-student framework, namely Multimodal Robust Prompt Distillation (MRPD) for distilling robust 3D point cloud model. It learns lightweight prompts by aligning student point cloud model's features with robust embeddings from three distinct teachers: a vision model processing depth projections, a high-performance 3D model, and a text encoder. To ensure a reliable knowledge transfer, this distillation is guided by a confidence-gated mechanism which dynamically balances the contribution of all input modalities. Notably, since the distillation is all during the training stage, there is no additional computational cost at inference. Extensive experiments demonstrate that MRPD substantially outperforms state-of-the-art defense methods against a wide range of white-box and black-box attacks, while even achieving better performance on clean data. Our work presents a new, practical paradigm for building robust 3D vision systems by efficiently harnessing multimodal knowledge.

Zixin Zhang, Chenfei Liao, Hongfei Zhang et al.

Affordance prediction serves as a critical bridge between perception and action in embodied AI. However, existing research is confined to pinhole camera models, which suffer from narrow Fields of View (FoV) and fragmented observations, often missing critical holistic environmental context. In this paper, we present the first exploration into Panoramic Affordance Prediction, utilizing 360-degree imagery to capture global spatial relationships and holistic scene understanding. To facilitate this novel task, we first introduce PAP-12K, a large-scale benchmark dataset containing over 1,000 ultra-high-resolution (12k, 11904 x 5952) panoramic images with over 12k carefully annotated QA pairs and affordance masks. Furthermore, we propose PAP, a training-free, coarse-to-fine pipeline inspired by the human foveal visual system to tackle the ultra-high resolution and severe distortion inherent in panoramic images. PAP employs recursive visual routing via grid prompting to progressively locate targets, applies an adaptive gaze mechanism to rectify local geometric distortions, and utilizes a cascaded grounding pipeline to extract precise instance-level masks. Experimental results on PAP-12K reveal that existing affordance prediction methods designed for standard perspective images suffer severe performance degradation and fail due to the unique challenges of panoramic vision. In contrast, PAP framework effectively overcomes these obstacles, significantly outperforming state-of-the-art baselines and highlighting the immense potential of panoramic perception for robust embodied intelligence.

Shouwei Gao, Xu Zheng, Dongsheng Luo et al.

The rapid growth of scientific machine learning (SciML) has accelerated discovery across diverse domains, yet designing effective SciML models remains a challenging task. In practice, building such models often requires substantial prior knowledge and manual expertise, particularly in determining which input features to use and how large the model should be. We introduce LUMOS, an end-to-end framework based on L0-regularized learning that unifies feature selection and model pruning to democratize SciML model design. By employing semi-stochastic gating and reparameterization techniques, LUMOS dynamically selects informative features and prunes redundant parameters during training, reducing the reliance on manual tuning while maintaining predictive accuracy. We evaluate LUMOS across 13 diverse SciML workloads, including cosmology and molecular sciences, and demonstrate its effectiveness and generalizability. Experiments on 13 SciML models show that LUMOS achieves 71.45% parameter reduction and a 6.4x inference speedup on average. Furthermore, Distributed Data Parallel (DDP) training on up to eight GPUs confirms the scalability of

Xu Zheng, Zihao Dongfang, Lutao Jiang et al.

Humans possess spatial reasoning abilities that enable them to understand spaces through multimodal observations, such as vision and sound. Large multimodal reasoning models extend these abilities by learning to perceive and reason, showing promising performance across diverse spatial tasks. However, systematic reviews and publicly available benchmarks for these models remain limited. In this survey, we provide a comprehensive review of multimodal spatial reasoning tasks with large models, categorizing recent progress in multimodal large language models (MLLMs) and introducing open benchmarks for evaluation. We begin by outlining general spatial reasoning, focusing on post-training techniques, explainability, and architecture. Beyond classical 2D tasks, we examine spatial relationship reasoning, scene and layout understanding, as well as visual question answering and grounding in 3D space. We also review advances in embodied AI, including vision-language navigation and action models. Additionally, we consider emerging modalities such as audio and egocentric video, which contribute to novel spatial understanding through new sensors. We believe this survey establishes a solid foundation and offers insights into the growing field of multimodal spatial reasoning. Updated information about this survey, codes and implementation of the open benchmarks can be found at https://github.com/zhengxuJosh/Awesome-Spatial-Reasoning.

Zixin Zhang, Kanghao Chen, Xingwang Lin et al.

The ability to use, understand, and create tools is a hallmark of human intelligence, enabling sophisticated interaction with the physical world. For any general-purpose intelligent agent to achieve true versatility, it must also master these fundamental skills. While modern Multimodal Large Language Models (MLLMs) leverage their extensive common knowledge for high-level planning in embodied AI and in downstream Vision-Language-Action (VLA) models, the extent of their true understanding of physical tools remains unquantified. To bridge this gap, we present PhysToolBench, the first benchmark dedicated to evaluating the comprehension of physical tools by MLLMs. Our benchmark is structured as a Visual Question Answering (VQA) dataset comprising over 1,000 image-text pairs. It assesses capabilities across three distinct difficulty levels: (1) Tool Recognition: Requiring the recognition of a tool's primary function. (2) Tool Understanding: Testing the ability to grasp the underlying principles of a tool's operation. (3) Tool Creation: Challenging the model to fashion a new tool from surrounding objects when conventional options are unavailable. Our comprehensive evaluation of 32 MLLMs-spanning proprietary, open-source, specialized embodied, and backbones in VLAs-reveals a significant deficiency in tool understanding. Furthermore, we provide an in-depth analysis and propose preliminary solutions. Code and dataset are publicly available.

Yihong Cao, Jiaming Zhang, Xu Zheng et al.

Panoramic image processing is essential for omni-context perception, yet faces constraints like distortions, perspective occlusions, and limited annotations. Previous unsupervised domain adaptation methods transfer knowledge from labeled pinhole data to unlabeled panoramic images, but they require access to source pinhole data. To address these, we introduce a more practical task, i.e., Source-Free Occlusion-Aware Seamless Segmentation (SFOASS), and propose its first solution, called UNconstrained Learning Omni-Context Knowledge (UNLOCK). Specifically, UNLOCK includes two key modules: Omni Pseudo-Labeling Learning and Amodal-Driven Context Learning. While adapting without relying on source data or target labels, this framework enhances models to achieve segmentation with 360° viewpoint coverage and occlusion-aware reasoning. Furthermore, we benchmark the proposed SFOASS task through both real-to-real and synthetic-to-real adaptation settings. Experimental results show that our source-free method achieves performance comparable to source-dependent methods, yielding state-of-the-art scores of 10.9 in mAAP and 11.6 in mAP, along with an absolute improvement of +4.3 in mAPQ over the source-only method. All data and code will be made publicly available at https://github.com/yihong-97/UNLOCK.

Bin Ren, Yawei Li, Xu Zheng et al.

Image Restoration (IR) aims to recover high quality images from degraded inputs affected by various corruptions such as noise, blur, haze, rain, and low light conditions. Despite recent advances, most existing approaches treat IR as a direct mapping problem, relying on shared representations across degradation types without modeling their structural diversity. In this work, we present MIRAGE, a unified and lightweight framework for all in one IR that explicitly decomposes the input feature space into three semantically aligned parallel branches, each processed by a specialized module attention for global context, convolution for local textures, and MLP for channel-wise statistics. This modular decomposition significantly improves generalization and efficiency across diverse degradations. Furthermore, we introduce a cross layer contrastive learning scheme that aligns shallow and latent features to enhance the discriminability of shared representations. To better capture the underlying geometry of feature representations, we perform contrastive learning in a Symmetric Positive Definite (SPD) manifold space rather than the conventional Euclidean space. Extensive experiments show that MIRAGE not only achieves new state of the art performance across a variety of degradation types but also offers a scalable solution for challenging all-in-one IR scenarios. Our code and models will be publicly available at https://amazingren.github.io/MIRAGE/.

Zhuomin Chen, Jingchao Ni, Hojat Allah Salehi et al.

Self-supervised graph representation learning (GRL) typically generates paired graph augmentations from each graph to infer similar representations for augmentations of the same graph, but distinguishable representations for different graphs. While effective augmentation requires both semantics-preservation and data-perturbation, most existing GRL methods focus solely on data-perturbation, leading to suboptimal solutions. To fill the gap, in this paper, we propose a novel method, Explanation-Preserving Augmentation (EPA), which leverages graph explanation for semantics-preservation. EPA first uses a small number of labels to train a graph explainer, which infers the subgraphs that explain the graph's label. Then these explanations are used for generating semantics-preserving augmentations for boosting self-supervised GRL. Thus, the entire process, namely EPA-GRL, is semi-supervised. We demonstrate theoretically, using an analytical example, and through extensive experiments on a variety of benchmark datasets, that EPA-GRL outperforms the state-of-the-art (SOTA) GRL methods that use semantics-agnostic augmentations. The code is available at https://github.com/realMoana/EPA-GRL.

Yuangong Chen, Wai Keung Wong, Jiaxing Li et al.

Multimodal Large Language Models (MLLMs) show strong visual perception, yet remain limited in reasoning about space under changing viewpoints. We study this challenge as Perspective-Conditioned Spatial Reasoning (PCSR) in 360-degree omnidirectional images, where broad scene coverage reduces ambiguity from partial observations without eliminating the need for viewpoint-dependent inference. To assess this capability, we introduce PCSR-Bench, a diagnostic benchmark of 84,373 question-answer pairs from 2,600 omnidirectional images across 26 indoor environments. PCSR-Bench contains eight tasks spanning foundational perception (e.g., object counting, relative distance, and relative direction) and advanced PCSR, including compositional chains, egocentric rotation, perspective re-anchoring, ego-distortion, and limited-FOV visibility. We evaluate 14 representative MLLMs and observe a substantial perception-reasoning gap: accuracy reaches 57.59% on foundational relative direction, but drops to 13.49% on egocentric rotation, 7.13% on egocentric distortion, and 0.64% on open-ended compositional reasoning. To probe the plasticity of this gap, we conduct an RL-based diagnostic study on a 7B-scale model. Reward shaping improves a matched 7B baseline from 31.10% to 60.06% under a controlled setting, suggesting that PCSR is partial plasticity rather than being fully immutable. Still, the gains are task-selective, sensitive to reward design including both weight allocation and reward formulation, and partially dependent on the evaluation protocol. These results position PCSR as a key bottleneck in current MLLMs and highlight limited but meaningful room for recovery under targeted optimization.

Yaowen Chang, Zhen Cao, Xu Zheng et al.

Panoramic semantic segmentation is pivotal for comprehensive 360° scene understanding in critical applications like autonomous driving and virtual reality. However, progress in this domain is constrained by two key challenges: the severe geometric distortions inherent in panoramic projections and the prohibitive cost of dense annotation. While Unsupervised Domain Adaptation (UDA) from label-rich pinhole-camera datasets offers a viable alternative, many real-world tasks impose a stricter source-free (SFUDA) constraint where source data is inaccessible for privacy or proprietary reasons. This constraint significantly amplifies the core problems of domain shift, leading to unreliable pseudo-labels and dramatic performance degradation, particularly for minority classes. To overcome these limitations, we propose the DAPASS framework. DAPASS introduces two synergistic modules to robustly transfer knowledge without source data. First, our Panoramic Confidence-Guided Denoising (PCGD) module generates high-fidelity, class-balanced pseudo-labels by enforcing perturbation consistency and incorporating neighborhood-level confidence to filter noise. Second, a Contextual Resolution Adversarial Module (CRAM) explicitly addresses scale variance and distortion by adversarially aligning fine-grained details from high-resolution crops with global semantics from low-resolution contexts. DAPASS achieves state-of-the-art performances on outdoor (Cityscapes-to-DensePASS) and indoor (Stanford2D3D) benchmarks, yielding 55.04% (+2.05%) and 70.38% (+1.54%) mIoU, respectively.

Chenfei Liao, Wensong Wang, Zichen Wen et al.

Recent endeavors to accelerate inference in Multimodal Large Language Models (MLLMs) have primarily focused on visual token compression. The effectiveness of these methods is typically assessed by measuring the accuracy drop on established benchmarks, comparing model performance before and after compression. However, these benchmarks are originally designed to assess the perception and reasoning capabilities of MLLMs, rather than to evaluate compression techniques. As a result, directly applying them to visual token compression introduces a task mismatch. Strikingly, our investigation reveals that simple image downsampling consistently outperforms many advanced compression methods across multiple widely used benchmarks. Through extensive experiments, we make the following observations: (i) Current benchmarks are noisy for the visual token compression task. (ii) Down-sampling is able to serve as a data filter to evaluate the difficulty of samples in the visual token compression task. Motivated by these findings, we introduce VTC-Bench, an evaluation framework that incorporates a data filtering mechanism to denoise existing benchmarks, thereby enabling fairer and more accurate assessment of visual token compression methods. All data and code are available at https://github.com/Chenfei-Liao/VTC-Bench.

Yuanhuiyi Lyu, Chi Kit Wong, Chenfei Liao et al.

Recent works have made notable advancements in enhancing unified models for text-to-image generation through the Chain-of-Thought (CoT). However, these reasoning methods separate the processes of understanding and generation, which limits their ability to guide the reasoning of unified models in addressing the deficiencies of their generative capabilities. To this end, we propose a novel reasoning framework for unified models, Understanding-in-Generation (UiG), which harnesses the robust understanding capabilities of unified models to reinforce their performance in image generation. The core insight of our UiG is to integrate generative guidance by the strong understanding capabilities during the reasoning process, thereby mitigating the limitations of generative abilities. To achieve this, we introduce "Image Editing" as a bridge to infuse understanding into the generation process. Initially, we verify the generated image and incorporate the understanding of unified models into the editing instructions. Subsequently, we enhance the generated image step by step, gradually infusing the understanding into the generation process. Our UiG framework demonstrates a significant performance improvement in text-to-image generation over existing text-to-image reasoning methods, e.g., a 3.92% gain on the long prompt setting of the TIIF benchmark. The project code: https://github.com/QC-LY/UiG

Liming Lu, Shuchao Pang, Xu Zheng et al.

Adversarial robustness distillation (ARD) aims to transfer both performance and robustness from teacher model to lightweight student model, enabling resilient performance on resource-constrained scenarios. Though existing ARD approaches enhance student model's robustness, the inevitable by-product leads to the degraded performance on clean examples. We summarize the causes of this problem inherent in existing methods with dual-teacher framework as: 1. The divergent optimization objectives of dual-teacher models, i.e., the clean and robust teachers, impede effective knowledge transfer to the student model, and 2. The iteratively generated adversarial examples during training lead to performance deterioration of the robust teacher model. To address these challenges, we propose a novel Cyclic Iterative ARD (CIARD) method with two key innovations: a. A multi-teacher framework with contrastive push-loss alignment to resolve conflicts in dual-teacher optimization objectives, and b. Continuous adversarial retraining to maintain dynamic teacher robustness against performance degradation from the varying adversarial examples. Extensive experiments on CIFAR-10, CIFAR-100, and Tiny-ImageNet demonstrate that CIARD achieves remarkable performance with an average 3.53 improvement in adversarial defense rates across various attack scenarios and a 5.87 increase in clean sample accuracy, establishing a new benchmark for balancing model robustness and generalization. Our code is available at https://github.com/eminentgu/CIARD

Zekai Lin, Xu Zheng

360 panoramic images are increasingly used in virtual reality, autonomous driving, and robotics for holistic scene understanding. However, current Vision-Language Models (VLMs) struggle with 3D spatial reasoning on Equirectangular Projection (ERP) images due to geometric distortion and limited 3D supervision. We introduce PanoEnv, a large-scale VQA benchmark built from synthetic 3D environments, containing 14.8K questions across five categories (e.g., relative position, volume comparison) grounded in accurate 3D annotations including depth, segmentation, and bounding boxes. Benchmarking 14 state-of-the-art VLMs reveals limited 3D understanding, achieving only 49.34% overall accuracy and 8.36% on open-ended (OE) questions. To enhance 3D reasoning, we propose a reinforcement learning post-training framework based on Group Relative Policy Optimization (GRPO) with a ground-truth-guided reward that incorporates five geometry-aware strategies such as distance tolerance and spatial consistency. A two-stage curriculum further mitigates catastrophic forgetting: Stage 1 trains on structured tasks (true/false and multiple choice), and Stage 2 fine-tunes on mixed open-ended data to improve generalization. Our 7B model achieves new state-of-the-art performance, improving overall accuracy to 52.93% (+3.59%) and open-ended accuracy to 14.83% while maintaining structured-task performance. It also achieves top semantic evaluation scores (Q-Score 6.24, P-Score 5.95), surpassing 32B models. These results demonstrate that PanoEnv-QA and our curriculum-based RL framework effectively instill 3D spatial intelligence in VLMs for omnidirectional perception.

Donghang Duan, Xu Zheng

Current LLM-based text anonymization frameworks usually rely on remote API services from powerful LLMs, which creates an inherent "privacy paradox": users must somehow disclose data to untrusted third parties for superior privacy preservation. Moreover, directly migrating these frameworks to local small-scale models (LSMs) offers a suboptimal solution with catastrophic collapse in utility based on our core findings. Our work argues that this failure stems not merely from the capability deficits of LSMs, but from the inherent irrationality of the greedy adversarial strategies employed by current state-of-the-art (SoTA) methods. We model the anonymization process as a trade-off between Marginal Privacy Gain (MPG) and Marginal Utility Cost (MUC), and demonstrate that greedy strategies inevitably drift into an irrational state. To address this, we propose Rational Localized Adversarial Anonymization (RLAA), a fully localized and training-free framework featuring an Attacker-Arbitrator-Anonymizer (A-A-A) architecture. RLAA introduces an arbitrator that acts as a rationality gatekeeper, validating the attacker's inference to filter out feedback providing negligible benefits on privacy preservation. This mechanism enforces a rational early-stopping criterion, and systematically prevents utility collapse. Extensive experiments on different datasets demonstrate that RLAA achieves the best privacy-utility trade-off, and in some cases even outperforms SoTA on the Pareto principle. Our code and datasets will be released upon acceptance.

Yibo Yan, Jiamin Su, Jianxiang He et al.

Mathematical reasoning, a core aspect of human cognition, is vital across many domains, from educational problem-solving to scientific advancements. As artificial general intelligence (AGI) progresses, integrating large language models (LLMs) with mathematical reasoning tasks is becoming increasingly significant. This survey provides the first comprehensive analysis of mathematical reasoning in the era of multimodal large language models (MLLMs). We review over 200 studies published since 2021, and examine the state-of-the-art developments in Math-LLMs, with a focus on multimodal settings. We categorize the field into three dimensions: benchmarks, methodologies, and challenges. In particular, we explore multimodal mathematical reasoning pipeline, as well as the role of (M)LLMs and the associated methodologies. Finally, we identify five major challenges hindering the realization of AGI in this domain, offering insights into the future direction for enhancing multimodal reasoning capabilities. This survey serves as a critical resource for the research community in advancing the capabilities of LLMs to tackle complex multimodal reasoning tasks.

Xu Zheng, Feiyu Wu, Zhuocheng Wang et al.

Language data are increasingly acquired and governed as assets, yet platforms often price candidate resources before knowing their true privacy or access costs. We study online pricing for governed language data assets under cost uncertainty. At each round, a platform observes an NLP task, a candidate asset, and a coarse cost estimate, may pay for a refined cost signal, posts a price, and receives safe net revenue. We introduce \textsc{NH-CROP}, a clipped robust pricing framework with a no-harm information-acquisition gate. The method compares direct pricing, risk-aware pricing, and verify-then-price, and acquires information only when its estimated decision value exceeds the best no-verification alternative. Across synthetic, real-proxy, and downstream-utility-grounded benchmarks, clipped \textsc{NH-CROP} variants improve or remain competitive with price-only and risk-aware baselines. Causal ablations show that paid verification is not the main source of gains in real-proxy and utility-grounded settings: the strongest learned policies often choose not to verify. Oracle and high-decision-value diagnostics show that refined cost information can still have substantial local value. Overall, governed language-data platforms should calibrate pricing under uncertain access costs first and verify only when information is cheap and decision-actionable.

Xu Zheng, Lin Wang

In this paper, we make the first attempt at achieving the cross-modal (i.e., image-to-events) adaptation for event-based object recognition without accessing any labeled source image data owning to privacy and commercial issues. Tackling this novel problem is non-trivial due to the novelty of event cameras and the distinct modality gap between images and events. In particular, as only the source model is available, a hurdle is how to extract the knowledge from the source model by only using the unlabeled target event data while achieving knowledge transfer. To this end, we propose a novel framework, dubbed EventDance for this unsupervised source-free cross-modal adaptation problem. Importantly, inspired by event-to-video reconstruction methods, we propose a reconstruction-based modality bridging (RMB) module, which reconstructs intensity frames from events in a self-supervised manner. This makes it possible to build up the surrogate images to extract the knowledge (i.e., labels) from the source model. We then propose a multi-representation knowledge adaptation (MKA) module that transfers the knowledge to target models learning events with multiple representation types for fully exploring the spatiotemporal information of events. The two modules connecting the source and target models are mutually updated so as to achieve the best performance. Experiments on three benchmark datasets with two adaption settings show that EventDance is on par with prior methods utilizing the source data.

Xu Zheng, Pengyuan Zhou, Athanasios V. Vasilakos et al.

In this paper, we address the challenging source-free unsupervised domain adaptation (SFUDA) for pinhole-to-panoramic semantic segmentation, given only a pinhole image pre-trained model (i.e., source) and unlabeled panoramic images (i.e., target). Tackling this problem is non-trivial due to three critical challenges: 1) semantic mismatches from the distinct Field-of-View (FoV) between domains, 2) style discrepancies inherent in the UDA problem, and 3) inevitable distortion of the panoramic images. To tackle these problems, we propose 360SFUDA++ that effectively extracts knowledge from the source pinhole model with only unlabeled panoramic images and transfers the reliable knowledge to the target panoramic domain. Specifically, we first utilize Tangent Projection (TP) as it has less distortion and meanwhile slits the equirectangular projection (ERP) to patches with fixed FoV projection (FFP) to mimic the pinhole images. Both projections are shown effective in extracting knowledge from the source model. However, as the distinct projections make it less possible to directly transfer knowledge between domains, we then propose Reliable Panoramic Prototype Adaptation Module (RP2AM) to transfer knowledge at both prediction and prototype levels. RP$^2$AM selects the confident knowledge and integrates panoramic prototypes for reliable knowledge adaptation. Moreover, we introduce Cross-projection Dual Attention Module (CDAM), which better aligns the spatial and channel characteristics across projections at the feature level between domains. Both knowledge extraction and transfer processes are synchronously updated to reach the best performance. Extensive experiments on the synthetic and real-world benchmarks, including outdoor and indoor scenarios, demonstrate that our 360SFUDA++ achieves significantly better performance than prior SFUDA methods.