arXiv subjects

Xinlong Chen

Publications and source records attributed to Xinlong Chen.

30 records · Page 2Linked to original sources

arXiv2026

Research on World Models Is Not Merely Injecting World Knowledge into Specific Tasks

World models have emerged as a critical frontier in AI research, aiming to enhance large models by infusing them with physical dynamics and world knowledge. The core objective is to enable agents to understand, predict, and interact with complex environments. However, current research landscape remains fragmented, with approaches predominantly focused on injecting world knowledge into isolated tasks, such as visual prediction, 3D estimation, or symbol grounding, rather than establishing a unified definition or framework. While these task-specific integrations yield performance gains, they often lack the systematic coherence required for holistic world understanding. In this paper, we analyze the limitations of such fragmented approaches and propose a unified design specification for world models. We suggest that a robust world model should not be a loose collection of capabilities but a normative framework that integrally incorporates interaction, perception, symbolic reasoning, and spatial representation. This work aims to provide a structured perspective to guide future research toward more general, robust, and principled models of the world.

cs.CV↗

arXiv2026

DiaDem: Advancing Dialogue Descriptions in Audiovisual Video Captioning for Multimodal Large Language Models

Accurate dialogue description in audiovisual video captioning is crucial for downstream understanding and generation tasks. However, existing models generally struggle to produce faithful dialogue descriptions within audiovisual captions. To mitigate this limitation, we propose DiaDem, a powerful audiovisual video captioning model capable of generating captions with more precise dialogue descriptions while maintaining strong overall performance. We first synthesize a high-quality dataset for SFT, then employ a difficulty-partitioned two-stage GRPO strategy to further enhance dialogue descriptions. To enable systematic evaluation of dialogue description capabilities, we introduce DiaDemBench, a comprehensive benchmark designed to evaluate models across diverse dialogue scenarios, emphasizing both speaker attribution accuracy and utterance transcription fidelity in audiovisual captions. Extensive experiments on DiaDemBench reveal even commercial models still exhibit substantial room for improvement in dialogue-aware captioning. Notably, DiaDem not only outperforms the Gemini series in dialogue description accuracy but also achieves competitive performance on general audiovisual captioning benchmarks, demonstrating its overall effectiveness.

cs.CL↗

arXiv2025

The Unseen Bias: How Norm Discrepancy in Pre-Norm MLLMs Leads to Visual Information Loss

Multimodal Large Language Models (MLLMs), which couple pre-trained vision encoders and language models, have shown remarkable capabilities. However, their reliance on the ubiquitous Pre-Norm architecture introduces a subtle yet critical flaw: a severe norm disparity between the high-norm visual tokens and the low-norm text tokens. In this work, we present a formal theoretical analysis demonstrating that this imbalance is not a static issue. Instead, it induces an ``asymmetric update dynamic,'' where high-norm visual tokens exhibit a ``representational inertia,'' causing them to transform semantically much slower than their textual counterparts. This fundamentally impairs effective cross-modal feature fusion. Our empirical validation across a range of mainstream MLLMs confirms that this theoretical dynamic -- the persistence of norm disparity and the resulting asymmetric update rates -- is a prevalent phenomenon. Based on this insight, we propose a remarkably simple yet effective solution: inserting a single, carefully initialized LayerNorm layer after the visual projector to enforce norm alignment. Experiments conducted on the LLaVA-1.5 architecture show that this intervention yields significant performance gains not only on a wide suite of multimodal benchmarks but also, notably, on text-only evaluations such as MMLU, suggesting that resolving the architectural imbalance leads to a more holistically capable model.

cs.CV↗

arXiv2025

Mavors: Multi-granularity Video Representation for Multimodal Large Language Model

Long-context video understanding in multimodal large language models (MLLMs) faces a critical challenge: balancing computational efficiency with the retention of fine-grained spatio-temporal patterns. Existing approaches (e.g., sparse sampling, dense sampling with low resolution, and token compression) suffer from significant information loss in temporal dynamics, spatial details, or subtle interactions, particularly in videos with complex motion or varying resolutions. To address this, we propose $\mathbf{Mavors}$, a novel framework that introduces $\mathbf{M}$ulti-gr$\mathbf{a}$nularity $\mathbf{v}$ide$\mathbf{o}$ $\mathbf{r}$epre$\mathbf{s}$entation for holistic long-video modeling. Specifically, Mavors directly encodes raw video content into latent representations through two core components: 1) an Intra-chunk Vision Encoder (IVE) that preserves high-resolution spatial features via 3D convolutions and Vision Transformers, and 2) an Inter-chunk Feature Aggregator (IFA) that establishes temporal coherence across chunks using transformer-based dependency modeling with chunk-level rotary position encodings. Moreover, the framework unifies image and video understanding by treating images as single-frame videos via sub-image decomposition. Experiments across diverse benchmarks demonstrate Mavors' superiority in maintaining both spatial fidelity and temporal continuity, significantly outperforming existing methods in tasks requiring fine-grained spatio-temporal reasoning.

cs.CV↗

arXiv2025

AVoCaDO: An Audiovisual Video Captioner Driven by Temporal Orchestration

Audiovisual video captioning aims to generate semantically rich descriptions with temporal alignment between visual and auditory events, thereby benefiting both video understanding and generation. In this paper, we present AVoCaDO, a powerful audiovisual video captioner driven by the temporal orchestration between audio and visual modalities. We propose a two-stage post-training pipeline: (1) AVoCaDO SFT, which fine-tunes the model on a newly curated dataset of 107K high-quality, temporally-aligned audiovisual captions; and (2) AVoCaDO GRPO, which leverages tailored reward functions to further enhance temporal coherence and dialogue accuracy while regularizing caption length and reducing collapse. Experimental results demonstrate that AVoCaDO significantly outperforms existing open-source models across four audiovisual video captioning benchmarks, and also achieves competitive performance on the VDC and DREAM-1K benchmark under visual-only settings.

cs.CV↗

arXiv2025

VidBridge-R1: Bridging QA and Captioning for RL-based Video Understanding Models with Intermediate Proxy Tasks

The "Reason-Then-Respond" paradigm, enhanced by Reinforcement Learning, has shown great promise in advancing Multimodal Large Language Models. However, its application to the video domain has led to specialized models that excel at either question answering (QA) or captioning tasks, but struggle to master both. Naively combining reward signals from these tasks results in mutual performance degradation, which we attribute to a conflict between their opposing task natures. To address this challenge, we propose a novel training framework built upon two intermediate proxy tasks: DarkEventInfer, which presents videos with masked event segments, requiring models to infer the obscured content based on contextual video cues; and MixVidQA, which presents interleaved video sequences composed of two distinct clips, challenging models to isolate and reason about one while disregarding the other. These proxy tasks compel the model to simultaneously develop both holistic, divergent understanding and precise, convergent reasoning capabilities. Embodying this framework, we present VidBridge-R1, the first versatile video reasoning model that effectively bridges the paradigm conflict. Extensive experiments show that VidBridge-R1 achieves significant performance gains on both QA and captioning within one model, demonstrating the efficacy of our approach in fostering more generalizable and powerful video understanding models.

cs.CV↗

arXiv2025

D$^2$HScore: Reasoning-Aware Hallucination Detection via Semantic Breadth and Depth Analysis in LLMs

Although large Language Models (LLMs) have achieved remarkable success, their practical application is often hindered by the generation of non-factual content, which is called "hallucination". Ensuring the reliability of LLMs' outputs is a critical challenge, particularly in high-stakes domains such as finance, security, and healthcare. In this work, we revisit hallucination detection from the perspective of model architecture and generation dynamics. Leveraging the multi-layer structure and autoregressive decoding process of LLMs, we decompose hallucination signals into two complementary dimensions: the semantic breadth of token representations within each layer, and the semantic depth of core concepts as they evolve across layers. Based on this insight, we propose \textbf{D$^2$HScore (Dispersion and Drift-based Hallucination Score)}, a training-free and label-free framework that jointly measures: (1) \textbf{Intra-Layer Dispersion}, which quantifies the semantic diversity of token representations within each layer; and (2) \textbf{Inter-Layer Drift}, which tracks the progressive transformation of key token representations across layers. To ensure drift reflects the evolution of meaningful semantics rather than noisy or redundant tokens, we guide token selection using attention signals. By capturing both the horizontal and vertical dynamics of representation during inference, D$^2$HScore provides an interpretable and lightweight proxy for hallucination detection. Extensive experiments across five open-source LLMs and five widely used benchmarks demonstrate that D$^2$HScore consistently outperforms existing training-free baselines.

cs.CL↗

arXiv2025

Attention-guided Self-reflection for Zero-shot Hallucination Detection in Large Language Models

Hallucination has emerged as a significant barrier to the effective application of Large Language Models (LLMs). In this work, we introduce a novel Attention-Guided SElf-Reflection (AGSER) approach for zero-shot hallucination detection in LLMs. The AGSER method utilizes attention contributions to categorize the input query into attentive and non-attentive queries. Each query is then processed separately through the LLMs, allowing us to compute consistency scores between the generated responses and the original answer. The difference between the two consistency scores serves as a hallucination estimator. In addition to its efficacy in detecting hallucinations, AGSER notably reduces computational overhead, requiring only three passes through the LLM and utilizing two sets of tokens. We have conducted extensive experiments with four widely-used LLMs across three different hallucination benchmarks, demonstrating that our approach significantly outperforms existing methods in zero-shot hallucination detection.

cs.CL↗

arXiv2025

Mixture of Decoding: An Attention-Inspired Adaptive Decoding Strategy to Mitigate Hallucinations in Large Vision-Language Models

Large Vision-Language Models (LVLMs) have exhibited impressive capabilities across various visual tasks, yet they remain hindered by the persistent challenge of hallucinations. To address this critical issue, we propose Mixture of Decoding (MoD), a novel approach for hallucination mitigation that dynamically adapts decoding strategies by evaluating the correctness of the model's attention on image tokens. Specifically, MoD measures the consistency between outputs generated from the original image tokens and those derived from the model's attended image tokens, to distinguish the correctness aforementioned. If the outputs are consistent, indicating correct attention, MoD employs a complementary strategy to amplify critical information. Conversely, if the outputs are inconsistent, suggesting erroneous attention, MoD utilizes a contrastive strategy to suppress misleading information. Extensive experiments demonstrate that MoD significantly outperforms existing decoding methods across multiple mainstream benchmarks, effectively mitigating hallucinations in LVLMs. The code is available at https://github.com/xlchen0205/MoD.

cs.CL↗

arXiv2025

VidCapBench: A Comprehensive Benchmark of Video Captioning for Controllable Text-to-Video Generation

The training of controllable text-to-video (T2V) models relies heavily on the alignment between videos and captions, yet little existing research connects video caption evaluation with T2V generation assessment. This paper introduces VidCapBench, a video caption evaluation scheme specifically designed for T2V generation, agnostic to any particular caption format. VidCapBench employs a data annotation pipeline, combining expert model labeling and human refinement, to associate each collected video with key information spanning video aesthetics, content, motion, and physical laws. VidCapBench then partitions these key information attributes into automatically assessable and manually assessable subsets, catering to both the rapid evaluation needs of agile development and the accuracy requirements of thorough validation. By evaluating numerous state-of-the-art captioning models, we demonstrate the superior stability and comprehensiveness of VidCapBench compared to existing video captioning evaluation approaches. Verification with off-the-shelf T2V models reveals a significant positive correlation between scores on VidCapBench and the T2V quality evaluation metrics, indicating that VidCapBench can provide valuable guidance for training T2V models. The project is available at https://github.com/VidCapBench/VidCapBench.

cs.AI↗

arXiv2023

Curriculum-Enhanced Residual Soft An-Isotropic Normalization for Over-smoothness in Deep GNNs

Despite Graph neural networks' significant performance gain over many classic techniques in various graph-related downstream tasks, their successes are restricted in shallow models due to over-smoothness and the difficulties of optimizations among many other issues. In this paper, to alleviate the over-smoothing issue, we propose a soft graph normalization method to preserve the diversities of node embeddings and prevent indiscrimination due to possible over-closeness. Combined with residual connections, we analyze the reason why the method can effectively capture the knowledge in both input graph structures and node features even with deep networks. Additionally, inspired by Curriculum Learning that learns easy examples before the hard ones, we propose a novel label-smoothing-based learning framework to enhance the optimization of deep GNNs, which iteratively smooths labels in an auxiliary graph and constructs many gradual non-smooth tasks for extracting increasingly complex knowledge and gradually discriminating nodes from coarse to fine. The method arguably reduces the risk of overfitting and generalizes better results. Finally, extensive experiments are carried out to demonstrate the effectiveness and potential of the proposed model and learning framework through comparison with twelve existing baselines including the state-of-the-art methods on twelve real-world node classification benchmarks.

cs.LG↗

arXiv2022

Spacecraft depth completion based on the gray image and the sparse depth map

Perceiving the three-dimensional (3D) structure of the spacecraft is a prerequisite for successfully executing many on-orbit space missions, and it can provide critical input for many downstream vision algorithms. In this paper, we propose to sense the 3D structure of spacecraft using light detection and ranging sensor (LIDAR) and a monocular camera. To this end, Spacecraft Depth Completion Network (SDCNet) is proposed to recover the dense depth map based on gray image and sparse depth map. Specifically, SDCNet decomposes the object-level spacecraft depth completion task into foreground segmentation subtask and foreground depth completion subtask, which segments the spacecraft region first and then performs depth completion on the segmented foreground area. In this way, the background interference to foreground spacecraft depth completion is effectively avoided. Moreover, an attention-based feature fusion module is also proposed to aggregate the complementary information between different inputs, which deduces the correlation between different features along the channel and the spatial dimension sequentially. Besides, four metrics are also proposed to evaluate object-level depth completion performance, which can more intuitively reflect the quality of spacecraft depth completion results. Finally, a large-scale satellite depth completion dataset is constructed for training and testing spacecraft depth completion algorithms. Empirical experiments on the dataset demonstrate the effectiveness of the proposed SDCNet, which achieves 0.25m mean absolute error of interest and 0.759m mean absolute truncation error, surpassing state-of-the-art methods by a large margin. The spacecraft pose estimation experiment is also conducted based on the depth completion results, and the experimental results indicate that the predicted dense depth map could meet the needs of downstream vision tasks.

cs.CV↗