arXiv subjects

Xilin Chen

Publications and source records attributed to Xilin Chen.

At least 19 recordsLinked to original sources

arXiv2026

EMPIRE: Explicit Manipulation Planning as a Learnable Intermediate Representation for Egocentric Hand-Motion Forecasting

Forecasting dexterous hand motions from egocentric observations is fundamental to intelligent interactive systems. Existing VLM-based methods typically map observations directly to future motions, overlooking the underlying manipulation process that governs hand-object interactions. Moreover, end-to-end optimization couples manipulation learning with motion synthesis, causing motion-generation gradients to interfere with the pre-learned manipulation-aware representations. To overcome these limitations, we propose EMPIRE, a two-stage framework that introduces Explicit Manipulation Planning as an Intermediate Representation for Egocentric hand-motion forecasting. Stage I: Learn to Plan. EMPIRE first learns explicit manipulation plans from multimodal context to capture the progression of hand-object interactions. Stage II: Learn to Act. A motion generator synthesizes future bimanual hand motions conditioned on frozen planner representations, preventing motion-generation gradients from affecting manipulation planning. To support our method, we further construct EMPIRE-651K, a bimanual hand-motion forecasting dataset comprising 650,910 training windows across 111 tasks, each paired with an explicit per-hand manipulation plan. Under identical training and evaluation protocols, EMPIRE achieves state-of-the-art forecasting accuracy, with an MPJPE of 84.53 mm and a finger-relative error of 38.97mm. We release the code and dataset at https://github.com/wangwen-banban/EMPIRE.

cs.RO↗

arXiv2026

Dual Adversarial Fine-tuning for Enhancing Robustness of Large Vision Language Model

While Large Vision-Language Models (LVLMs), represented by LLaVA and GPT-4V, have demonstrated remarkable capabilities, their visual inputs remain vulnerable to adversarial attacks, posing significant security risks. Existing defense methods predominantly target single-task scenarios (e.g., zero-shot classification) and consequently lack generalizability across various multimodal tasks. To address this limitation, we propose a dual adversarial fine-tuning framework that jointly optimizes visual and semantic supervision signals from two modalities, enhancing model robustness while generalizing across multiple downstream tasks. The proposed framework comprises two core components, i.e., $\textbf{Visual}$ supervision branch and $\textbf{Semantic}$ supervision branch. The former branch leverages features from clean images, extracted via a frozen original vision encoder, to guide adversarial robustness while the latter incorporates caption-image alignment as a contextual signal to preserve semantic coherence under attack. Moreover, our method achieves cross-task robustness by simply replacing the CLIP vision encoder in the original model, with no need of separate task-specific retraining or architecture modifications.Extensive experiments demonstrate that our approach outperforms the state-of-the-art method in adversarial robustness evaluation across zero-shot classification, image captioning, and visual question answering (VQA) tasks.

cs.CV↗

arXiv2026

DeepBias: Adaptive In-depth Probing of Social Biases in LVLMs

While Large Vision-Language Models (LVLMs) demonstrate remarkable capabilities, they remain highly susceptible to embedded social biases. Existing bias evaluation protocols predominantly rely on static datasets, which provide only a superficial assessment, as their fixed test cases cannot adaptively evolve to measure the true depth and limits of model vulnerabilities. We introduce DeepBias, an adaptive framework for the in-depth probing of social biases in LVLMs with carefully designed agents. Our approach operates through a dynamic ''generation-evolution-probing'' loop. First, a generative ProposerAgent synthesizes test data and is iteratively updated via Direct Preference Optimization (DPO) based on the target LVLM's responses, exploring model-specific failure modes. Second, an autonomous skill-driven DiggerAgent rewrites each test data across multiple probing turns, adaptively selecting from a curated skill library of deepening and rewriting strategies. At each turn, this process is conditioned on the model's previous response, enabling progressively deeper biases to be exposed. Furthermore, we build a benchmark named DeepBiasBench using our framework. By employing an ensemble of five diverse state-of-the-art LVLMs as anchors, the benchmark captures vulnerabilities shared across architectures. Comprehensive experiments demonstrate the effectiveness of our framework and show that DeepBias provides a challenging benchmark for in-depth bias evaluation, establishing an evolutionary paradigm for LVLM safety assessment.

cs.CY↗

arXiv2026

EntropyScan: Towards Model-level Backdoor Detection in LVLMs via Visual Attention Entropy

Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities across various tasks, yet they remain vulnerable to backdoor attacks. Existing defense methods predominantly focus on sample-level defense, which relies on the knowledge of training data or triggers. However, identifying whether a given model is backdoored remains a critical but unexplored task. To fill this gap, we propose EntropyScan, a lightweight and trigger-agnostic method for model-level backdoor detection in LVLMs. We first observe that backdoor injection disrupts the cross-modal alignment, resulting in pronounced structural anomalies in visual attention allocation on benign samples. Based on this insight, EntropyScan detects the backdoor models by quantifying such attention deviations. Specifically, it extracts visual attention distributions from the initial layers of the Large Language Model (LLM) and applies Tsallis entropy to capture these structural distortions. By employing a reference-anchored Z-score normalization on a small set of benign samples, it effectively identifies the backdoored model. Extensive experiments across two LVLMs architectures and three advanced attack scenarios show that EntropyScan achieves an F1 score of 98.5% in average and an AUC of 96.6%. Our code will be publicly available soon.

cs.CV↗

arXiv2026

Component-Based Out-of-Distribution Detection

Out-of-Distribution (OOD) detection requires sensitivity to subtle shifts without overreacting to natural In-Distribution (ID) diversity. However, from the viewpoint of detection granularity, global representation inevitably suppress local OOD cues, while patch-based methods are unstable due to entangled spurious-correlation and noise. And neither them is effective in detecting compositional OODs composed of valid ID components. Inspired by recognition-by-components theory, we present a training-free Component-Based OOD Detection (CoOD) framework that addresses the existing limitations by decomposing inputs into functional components. To instantiate CoOD, we derive Component Shift Score (CSS) to detect local appearance shifts, and Compositional Consistency Score (CCS) to identify cross-component compositional inconsistencies. Empirically, CoOD achieves consistent improvements on both coarse- and fine-grained OOD detection.

cs.CV↗

arXiv2026

EgoMotion: Hierarchical Reasoning and Diffusion for Egocentric Vision-Language Motion Generation

Faithfully modeling human behavior in dynamic environments is a foundational challenge for embodied intelligence. While conditional motion synthesis has achieved significant advances, egocentric motion generation remains largely underexplored due to the inherent complexity of first-person perception. In this work, we investigate Egocentric Vision-Language (Ego-VL) motion generation. This task requires synthesizing 3D human motion conditioned jointly on first-person visual observations and natural language instructions. We identify a critical \textit{reasoning-generation entanglement} challenge: the simultaneous optimization of semantic reasoning and kinematic modeling introduces gradient conflicts. These conflicts systematically degrade the fidelity of multimodal grounding and motion quality. To address this challenge, we propose a hierarchical generative framework \textbf{EgoMotion}. Inspired by the biological decoupling of cognitive reasoning and motor control, EgoMotion operates in two stages. In the Cognitive Reasoning stage, A vision-language model (VLM) projects multimodal inputs into a structured space of discrete motion primitives. This forces the VLM to acquire goal-consistent representations, effectively bridging the semantic gap between high-level perceptual understanding and low-level action execution. In the Motion Generation stage, these learned representations serve as expressive conditioning signals for a diffusion-based motion generator. By performing iterative denoising within a continuous latent space, the generator synthesizes physically plausible and temporally coherent trajectories. Extensive evaluations demonstrate that EgoMotion achieves state-of-the-art performance, and produces motion sequences that are both semantically grounded and kinematically superior to existing approaches.

cs.CV↗

arXiv2026

GEAR: GEometry-motion Alternating Refinement for Articulated Object Modeling with Gaussian Splatting

High-fidelity interactive digital assets are essential for embodied intelligence and robotic interaction, yet articulated objects remain challenging to reconstruct due to their complex structures and coupled geometry-motion relationships. Existing methods suffer from instability in geometry-motion joint optimization, while their generalization remains limited on complex multi-joint or out-of-distribution objects. To address these challenges, we propose GEAR, an EM-style alternating optimization framework that jointly models geometry and motion as interdependent components within a Gaussian Splatting representation. GEAR treats part segmentation as a latent variable and joint motion parameters as explicit variables, alternately refining them for improved convergence and geometric-motion consistency. To enhance part segmentation quality without sacrificing generalization, we leverage a vanilla 2D segmentation model to provide multi-view part priors, and employ a weakly supervised constraint to regularize the latent variable. Experiments on multiple benchmarks and our newly constructed dataset GEAR-Multi demonstrate that GEAR achieves state-of-the-art results in geometric reconstruction and motion parameters estimation, particularly on complex articulated objects with multiple movable parts.

cs.CV↗

arXiv2026

ACT Now: Preempting LVLM Hallucinations via Adaptive Context Integration

Large Vision-Language Models (LVLMs) frequently suffer from severe hallucination issues. Existing mitigation strategies predominantly rely on isolated, single-step states to enhance visual focus or suppress strong linguistic priors. However, these static approaches neglect dynamic context changes across the generation process and struggles to correct inherited information loss. To address this limitation, we propose Adaptive Context inTegration (ACT), a training-free inference intervention method that mitigates hallucination through the adaptive integration of contextual information. Specifically, we first propose visual context exploration, which leverages spatio-temporal profiling to adaptively amplify attention heads responsible for visual exploration. To further facilitate vision-language alignment, we propose semantic context aggregation that marginalizes potential semantic queries to effectively aggregate visual evidence, thereby resolving the information loss caused by the discrete nature of token prediction. Extensive experiments across diverse LVLMs demonstrate that ACT significantly reduces hallucinations and achieves competitive results on both discriminative and generative benchmarks, acting as a robust and highly adaptable solution without compromising fundamental generation capabilities.

cs.CV↗

arXiv2026

Neural Gate: Mitigating Privacy Risks in LVLMs via Neuron-Level Gradient Gating

Large Vision-Language Models (LVLMs) have shown remarkable potential across a wide array of vision-language tasks, leading to their adoption in critical domains such as finance and healthcare. However, their growing deployment also introduces significant security and privacy risks. Malicious actors could potentially exploit these models to extract sensitive information, highlighting a critical vulnerability. Recent studies show that LVLMs often fail to consistently refuse instructions designed to compromise user privacy. While existing work on privacy protection has made meaningful progress in preventing the leakage of sensitive data, they are constrained by limitations in both generalization and non-destructiveness. They often struggle to robustly handle unseen privacy-related queries and may inadvertently degrade a model's performance on standard tasks. To address these challenges, we introduce Neural Gate, a novel method for mitigating privacy risks through neuron-level model editing. Our method improves a model's privacy safeguards by increasing its rate of refusal for privacy-related questions, crucially extending this protective behavior to novel sensitive queries not encountered during the editing process. Neural Gate operates by learning a feature vector to identify neurons associated with privacy-related concepts within the model's representation of a subject. This localization then precisely guides the update of model parameters. Through comprehensive experiments on MiniGPT and LLaVA, we demonstrate that our method significantly boosts the model's privacy protection while preserving its original utility. The code is available at https://github.com/Xiangkui-Cao/Neural-Gate.

cs.CV↗

arXiv2026

What Makes VLMs Robust? Towards Reconciling Robustness and Accuracy in Vision-Language Models

Achieving adversarial robustness in Vision-Language Models (VLMs) inevitably compromises accuracy on clean data, presenting a long-standing and challenging trade-off. In this work, we revisit this trade-off by investigating a fundamental question: What makes VLMs robust? Through a detailed analysis of adversarially fine-tuned models, we examine how robustness mechanisms function internally and how they interact with clean accuracy. Our analysis reveals that adversarial robustness is not uniformly distributed across network depth. Instead, unexpectedly, it is primarily localized within the shallow layers, driven by a low-frequency spectral bias and input-insensitive attention patterns. Meanwhile, updates to the deep layers tend to undermine both clean accuracy and robust generalization. Motivated by these insights, we propose Adversarial Robustness Adaptation (R-Adapt), a simple yet effective framework that freezes all pre-trained weights and introduces minimal, insight-driven adaptations only in the initial layers. This design achieves an exceptional balance between adversarial robustness and clean accuracy. R-Adapt further supports training-free, model-guided, and data-driven paradigms, offering flexible pathways to seamlessly equip standard models with robustness. Extensive evaluations on 18 datasets and diverse tasks demonstrate our state-of-the-art performance under various attacks. Notably, R-Adapt generalizes efficiently to large vision-language models (e.g., LLaVA and Qwen-VL) to enhance their robustness. Our project page is available at https://summu77.github.io/R-Adapt.

cs.CV↗

arXiv2026

INFACT: A Diagnostic Benchmark for Induced Faithfulness and Factuality Hallucinations in Video-LLMs

Despite rapid progress, Video Large Language Models (Video-LLMs) remain unreliable due to hallucinations, which are outputs that contradict either video evidence (faithfulness) or verifiable world knowledge (factuality). Existing benchmarks provide limited coverage of factuality hallucinations and predominantly evaluate models only in clean settings. We introduce \textsc{INFACT}, a diagnostic benchmark comprising 9{,}800 QA instances with fine-grained taxonomies for faithfulness and factuality, spanning real and synthetic videos. \textsc{INFACT} evaluates models in four modes: Base (clean), Visual Degradation, Evidence Corruption, and Temporal Intervention for order-sensitive items. Reliability under induced modes is quantified using Resist Rate (RR) and Temporal Sensitivity Score (TSS). Experiments on 14 representative Video-LLMs reveal that higher Base-mode accuracy does not reliably translate to higher reliability in the induced modes, with evidence corruption reducing stability and temporal intervention yielding the largest degradation. Notably, many open-source baselines exhibit near-zero TSS on factuality, indicating pronounced temporal inertia on order-sensitive questions.

cs.CV↗

arXiv2026

From Semantics to Pixels: Coarse-to-Fine Masked Autoencoders for Hierarchical Visual Understanding

Self-supervised visual pre-training methods face an inherent tension: contrastive learning (CL) captures global semantics but loses fine-grained detail, while masked image modeling (MIM) preserves local textures but suffers from "attention drift" due to semantically-agnostic random masking. We propose C2FMAE, a coarse-to-fine masked autoencoder that resolves this tension by explicitly learning hierarchical visual representations across three data granularities: semantic masks (scene-level), instance masks (object-level), and RGB images (pixel-level). Two synergistic innovations enforce a strict top-down learning principle. First, a cascaded decoder sequentially reconstructs from scene semantics to object instances to pixel details, establishing explicit cross-granularity dependencies that parallel decoders cannot capture. Second, a progressive masking curriculum dynamically shifts the training focus from semantic-guided to instance-guided and finally to random masking, creating a structured learning path from global context to local features. To support this framework, we construct a large-scale multi-granular dataset with high-quality pseudo-labels for all 1.28M ImageNet-1K images. Extensive experiments show that C2FMAE achieves significant performance gains on image classification, object detection, and semantic segmentation, validating the effectiveness of our hierarchical design in learning more robust and generalizable representations.

cs.CV↗

arXiv2026

RoboPCA: Pose-centered Affordance Learning from Human Demonstrations for Robot Manipulation

Understanding spatial affordances -- comprising the contact regions of object interaction and the corresponding contact poses -- is essential for robots to effectively manipulate objects and accomplish diverse tasks. However, existing spatial affordance prediction methods mainly focus on locating the contact regions while delegating the pose to independent pose estimation approaches, which can lead to task failures due to inconsistencies between predicted contact regions and candidate poses. In this work, we propose RoboPCA, a pose-centered affordance prediction framework that jointly predicts task-appropriate contact regions and poses conditioned on instructions. To enable scalable data collection for pose-centered affordance learning, we devise Human2Afford, a data curation pipeline that automatically recovers scene-level 3D information and infers pose-centered affordance annotations from human demonstrations. With Human2Afford, scene depth and the interaction object's mask are extracted to provide 3D context and object localization, while pose-centered affordance annotations are obtained by tracking object points within the contact region and analyzing hand-object interaction patterns to establish a mapping from the 3D hand mesh to the robot end-effector orientation. By integrating geometry-appearance cues through an RGB-D encoder and incorporating mask-enhanced features to emphasize task-relevant object regions into the diffusion-based framework, RoboPCA outperforms baseline methods on image datasets, simulation, and real robots, and exhibits strong generalization across tasks and categories.

cs.RO↗

arXiv2026

Uni-Skill: Building Self-Evolving Skill Repository for Generalizable Robotic Manipulation

While skill-centric approaches leverage foundation models to enhance generalization in compositional tasks, they often rely on fixed skill libraries, limiting adaptability to new tasks without manual intervention. To address this, we propose Uni-Skill, a Unified Skill-centric framework that supports skill-aware planning and facilitates automatic skill evolution. Unlike prior methods that restrict planning to predefined skills, Uni-Skill requests for new skill implementations when existing ones are insufficient, ensuring adaptable planning with self-augmented skill library. To support automatic implementation of diverse skills requested by the planning module, we construct SkillFolder, a VerbNet-inspired repository derived from large-scale unstructured robotic videos. SkillFolder introduces a hierarchical skill taxonomy that captures diverse skill descriptions at multiple levels of abstraction. By populating this taxonomy with large-scale, automatically annotated demonstrations, Uni-Skill shifts the paradigm of skill acquisition from inefficient manual annotation to efficient offline structural retrieval. Retrieved examples provide semantic supervision over behavior patterns and fine-grained references for spatial trajectories, enabling few-shot skill inference without deployment-time demonstrations. Comprehensive experiments in both simulation and real-world settings verify the state-of-the-art performance of Uni-Skill over existing VLM-based skill-centric approaches, highlighting its advanced reasoning capabilities and strong zero-shot generalization across a wide range of novel tasks.

cs.RO↗

arXiv2026

Contrastive Spectral Rectification: Test-Time Defense towards Zero-shot Adversarial Robustness of CLIP

Vision-language models (VLMs) such as CLIP have demonstrated remarkable zero-shot generalization, yet remain highly vulnerable to adversarial examples (AEs). While test-time defenses are promising, existing methods fail to provide sufficient robustness against strong attacks and are often hampered by high inference latency and task-specific applicability. To address these limitations, we start by investigating the intrinsic properties of AEs, which reveals that AEs exhibit severe feature inconsistency under progressive frequency attenuation. We further attribute this to the model's inherent spectral bias. Leveraging this insight, we propose an efficient test-time defense named Contrastive Spectral Rectification (CSR). CSR optimizes a rectification perturbation to realign the input with the natural manifold under a spectral-guided contrastive objective, which is applied input-adaptively. Extensive experiments across 16 classification benchmarks demonstrate that CSR outperforms the SOTA by an average of 18.1% against strong APGD with modest inference overhead. Furthermore, CSR exhibits broad applicability across diverse visual tasks. Code is available at https://github.com/Summu77/CSR.

cs.CV↗

arXiv2025

T2VAttack: Adversarial Attack on Text-to-Video Diffusion Models

The rapid evolution of Text-to-Video (T2V) diffusion models has driven remarkable advancements in generating high-quality, temporally coherent videos from natural language descriptions. Despite these achievements, their vulnerability to adversarial attacks remains largely unexplored. In this paper, we introduce T2VAttack, a comprehensive study of adversarial attacks on T2V diffusion models from both semantic and temporal perspectives. Considering the inherently dynamic nature of video data, we propose two distinct attack objectives: a semantic objective to evaluate video-text alignment and a temporal objective to assess the temporal dynamics. To achieve an effective and efficient attack process, we propose two adversarial attack methods: (i) T2VAttack-S, which identifies semantically or temporally critical words in prompts and replaces them with synonyms via greedy search, and (ii) T2VAttack-I, which iteratively inserts optimized words with minimal perturbation to the prompt. By combining these objectives and strategies, we conduct a comprehensive evaluation on the adversarial robustness of several state-of-the-art T2V models, including ModelScope, CogVideoX, Open-Sora, and HunyuanVideo. Our experiments reveal that even minor prompt modifications, such as the substitution or insertion of a single word, can cause substantial degradation in semantic fidelity and temporal dynamics, highlighting critical vulnerabilities in current T2V diffusion models.

cs.CV↗

arXiv2025

Semantic Mismatch and Perceptual Degradation: A New Perspective on Image Editing Immunity

Text-guided image editing via diffusion models, while powerful, raises significant concerns about misuse, motivating efforts to immunize images against unauthorized edits using imperceptible perturbations. Prevailing metrics for evaluating immunization success typically rely on measuring the visual dissimilarity between the output generated from a protected image and a reference output generated from the unprotected original. This approach fundamentally overlooks the core requirement of image immunization, which is to disrupt semantic alignment with attacker intent, regardless of deviation from any specific output. We argue that immunization success should instead be defined by the edited output either semantically mismatching the prompt or suffering substantial perceptual degradations, both of which thwart malicious intent. To operationalize this principle, we propose Synergistic Intermediate Feature Manipulation (SIFM), a method that strategically perturbs intermediate diffusion features through dual synergistic objectives: (1) maximizing feature divergence from the original edit trajectory to disrupt semantic alignment with the expected edit, and (2) minimizing feature norms to induce perceptual degradations. Furthermore, we introduce the Immunization Success Rate (ISR), a novel metric designed to rigorously quantify true immunization efficacy for the first time. ISR quantifies the proportion of edits where immunization induces either semantic failure relative to the prompt or significant perceptual degradations, assessed via Multimodal Large Language Models (MLLMs). Extensive experiments show our SIFM achieves the state-of-the-art performance for safeguarding visual content against malicious diffusion-based manipulation.

cs.CV↗

arXiv2025

Dual Attention Guided Defense Against Malicious Edits

Recent progress in text-to-image diffusion models has transformed image editing via text prompts, yet this also introduces significant ethical challenges from potential misuse in creating deceptive or harmful content. While current defenses seek to mitigate this risk by embedding imperceptible perturbations, their effectiveness is limited against malicious tampering. To address this issue, we propose a Dual Attention-Guided Noise Perturbation (DANP) immunization method that adds imperceptible perturbations to disrupt the model's semantic understanding and generation process. DANP functions over multiple timesteps to manipulate both cross-attention maps and the noise prediction process, using a dynamic threshold to generate masks that identify text-relevant and irrelevant regions. It then reduces attention in relevant areas while increasing it in irrelevant ones, thereby misguides the edit towards incorrect regions and preserves the intended targets. Additionally, our method maximizes the discrepancy between the injected noise and the model's predicted noise to further interfere with the generation. By targeting both attention and noise prediction mechanisms, DANP exhibits impressive immunity against malicious edits, and extensive experiments confirm that our method achieves state-of-the-art performance.

cs.CV↗