# New Papers (Pending Classification) > Auto-generated by `arxiv_search.py`. Please review and classify into the appropriate sections in `README.md`. ## Update — 2026-05-01 **13 new paper(s) found.** ### [DualFact+: A Multimodal Fact Verification Framework for Procedural Video Understanding](https://arxiv.org/abs/2604.25584) - **arXiv ID:** `2604.25584` - **Date:** 04/2026 - **Code:** - - **Abstract:** We introduce DualFact, a dual-layer, multimodal factuality evaluation framework for procedural video captioning. DualFact separates factual correctness into conceptual facts, capturing abstract semantic roles (e.g., Action, Ingredient, Tool, Location), and contextual facts, capturing their grounded predicate-argument realizations in video. To support complete and role-consistent evaluation, DualFact incorporates implicit argument augmentation (VIA) and contrastive fact sets. We instantiate DualFact in two modes: DualFact-T, which verifies facts against textual evidence, and DualFact-V, which verifies facts against video-grounded visual evidence. Experiments on YouCook3-Fact and CraftBench-Fact show that state-of-the-art multimodal language models produce fluent but often factually incomplete captions, with systematic omissions and role-level inconsistencies. DualFact correlates more strongly with human factuality judgments than standard metrics, particularly for contextual facts, and reveals that caption-only evaluation overestimates **`hallucinations`** compared to video-grounded verification. Overall, DualFact offers an interpretable and human-aligned evaluation protocol that highlights persistent challenges in multimodal factual grounding, extending beyond surface-level fluency. --- ### [IntentVLM: Open-Vocabulary Intention Recognition through Forward-Inverse Modeling with Video-Language Models](https://arxiv.org/abs/2604.24002) - **arXiv ID:** `2604.24002` - **Date:** 04/2026 - **Code:** - - **Abstract:** Improving the effectiveness of human-robot interaction requires social robots to accurately infer human goals through robust intention understanding. This challenge is particularly critical in multimodal settings, where agents must integrate heterogeneous signals including text, visual cues to form a coherent interpretation of user intent. This paper presents IntentVLM, a novel two-stage video-language framework designed for open-vocabulary human intention recognition. The approach is inspired by forward-inverse modeling in cognitive science by decomposing intention understanding into goal candidate generation followed by structured inference through selection, effectively reducing **`hallucinations`** in latent reasoning. Evaluated on the IntentQA and Inst-IT Bench datasets, IntentVLM achieves state-of-the-art results with up to 80% accuracy, notably surpassing the baseline performance by 30% and matches human performance. Our findings demonstrate that this structured reasoning approach enhances open-vocabulary intention understanding without catastrophic forgetting, offering a robust foundation for human-centered robotics. --- ### [Exploring Audio Hallucination in Egocentric Video Understanding](https://arxiv.org/abs/2604.23860) - **arXiv ID:** `2604.23860` - **Date:** 04/2026 - **Code:** - - **Abstract:** Egocentric videos provide a distinctive setting in which sound serves as crucial cues to understand user activities and surroundings, particularly when visual information is unstable or occluded due to continuous camera movement. State-of-the-art large audio-visual language models (AV-LLMs) can generate multimodal descriptions. However, we show in this work that they are prone to audio **`hallucinations`**, often inferring sounds from visual cues that are visible but not heard. We present a systematic and automatic evaluation framework for analyzing audio **`hallucinations`** in egocentric video through a targeted question-answering (Q/A) protocol. We curate a dataset of 300 egocentric videos and design 1,000 sound-focused questions to probe model outputs. To characterize **`hallucinations`**, we propose a grounded taxonomy that distinguishes between foreground action sounds from the user activities and background ambient sounds. Our evaluation shows that advanced AV-LLMs, such as Qwen2.5 Omni, exhibit high **`hallucination`** rates, achieving only 27.3% and 39.5% accuracy on Q/As related to foreground and background sounds, respectively. With this work, we highlight the need to measure the reliability of multimodal responses, emphasizing that robust evaluation of **`hallucinations`** is essential to develop reliable AV-LLMs. --- ### [Sink-Token-Aware Pruning for Fine-Grained Video Understanding in Efficient Video LLMs](https://arxiv.org/abs/2604.20937) - **arXiv ID:** `2604.20937` - **Date:** 04/2026 - **Code:** - - **Abstract:** Video Large Language Models (Video LLMs) incur high inference latency due to a large number of visual tokens provided to LLMs. To address this, training-free visual token pruning has emerged as a solution to reduce computational costs; however, existing methods are primarily validated on Multiple-Choice Question Answering (MCQA) benchmarks, where coarse-grained cues often suffice. In this work, we reveal that these methods suffer a sharp performance collapse on fine-grained understanding tasks requiring precise visual grounding, such as **`hallucination`** evaluation. To explore this gap, we conduct a systematic analysis and identify sink tokens--semantically uninformative tokens that attract excessive attention--as a key obstacle to fine-grained video understanding. When these sink tokens survive pruning, they distort the model's visual evidence and hinder fine-grained understanding. Motivated by these insights, we propose Sink-Token-aware Pruning (SToP), a simple yet effective plug-and-play method that introduces a sink score to quantify each token's tendency to behave as a sink and applies this score to existing spatial and temporal pruning methods to suppress them, thereby enhancing video understanding. To validate the effectiveness of SToP, we apply it to state-of-the-art pruning methods (VisionZip, FastVid, and Holitom) and evaluate it across diverse benchmarks covering **`hallucination`**, open-ended generation, compositional reasoning, and MCQA. Our results demonstrate that SToP significantly boosts performance, even when pruning up to 90% of visual tokens. --- ### [Video-ToC: Video Tree-of-Cue Reasoning](https://arxiv.org/abs/2604.20473) - **arXiv ID:** `2604.20473` - **Date:** 04/2026 - **Code:** [Code](https://github.com/qizhongtan/Video-ToC) - **Abstract:** Existing Video Large Language Models (Video LLMs) struggle with complex video understanding, exhibiting limited reasoning capabilities and potential **`hallucinations`**. In particular, these methods tend to perform reasoning solely relying on the pretrained inherent reasoning rationales whilst lacking perception-aware adaptation to the input video content. To address this, we propose \textbf{Video-ToC}, a novel video reasoning framework that enhances video understanding through tree-of-cue reasoning. Specifically, our approach introduces three key innovations: (1) A tree-guided visual cue localization mechanism, which endows the model with enhanced fine-grained perceptual capabilities through structured reasoning patterns; (2) A reasoning-demand reward mechanism, which dynamically adjusts the reward value for reinforcement learning (RL) based on the estimation of reasoning demands, enabling on-demand incentives for more effective reasoning strategies; and (3) An automated annotation pipeline that constructs the Video-ToC-SFT-1k and Video-ToC-RL-2k datasets for supervised fine-tuning (SFT) and RL training, respectively. Extensive evaluations on six video understanding benchmarks and a video **`hallucination`** benchmark demonstrate the superiority of Video-ToC over baselines and recent methods. Code is available at https://github.com/qizhongtan/Video-ToC. --- ### [CCTVBench: Contrastive Consistency Traffic VideoQA Benchmark for Multimodal LLMs](https://arxiv.org/abs/2604.20460) - **arXiv ID:** `2604.20460` - **Date:** 04/2026 - **Code:** - - **Abstract:** Safety-critical traffic reasoning requires contrastive consistency: models must detect true hazards when an accident occurs, and reliably reject plausible-but-false hypotheses under near-identical counterfactual scenes. We present CCTVBench, a Contrastive Consistency Traffic VideoQA Benchmark built on paired real accident videos and world-model-generated counterfactual counterparts, together with minimally different, mutually exclusive hypothesis questions. CCTVBench enforces a single structured decision pattern over each video question quadruple and provides actionable diagnostics that decompose failures into positive omission, positive swap, negative **`hallucination`**, and mutual-exclusivity violation, while separating video versus question consistency. Experiments across open-source and proprietary video LLMs reveal a large and persistent gap between standard per-instance QA metrics and quadruple-level contrastive consistency, with unreliable none-of-the-above rejection as a key bottleneck. Finally, we introduce C-TCD, a contrastive decoding approach leveraging a semantically exclusive counterpart video as the contrast input at inference time, improving both instance-level QA and contrastive consistency. --- ### [Spatiotemporal Sycophancy: Negation-Based Gaslighting in Video Large Language Models](https://arxiv.org/abs/2604.17873) - **arXiv ID:** `2604.17873` - **Date:** 04/2026 - **Code:** - - **Abstract:** Video Large Language Models (Vid-LLMs) have demonstrated remarkable performance in video understanding tasks, yet their robustness under conversational interaction remains largely underexplored. In this paper, we identify spatiotemporal sycophancy, a failure mode in which Vid-LLMs retract initially correct, visually grounded judgments and conform to misleading user feedback under negation-based gaslighting. Rather than merely changing their answers, the models often fabricate unsupported temporal or spatial explanations to justify incorrect revisions. To systematically investigate this phenomenon, we propose a negation-based gaslighting evaluation framework and introduce GasVideo-1000, a curated benchmark designed to probe spatiotemporal sycophancy with clear visual grounding and temporal reasoning requirements. We evaluate a broad range of state-of-the-art open-source and proprietary Vid-LLMs across diverse video understanding tasks. Extensive experiments reveal that vulnerability to negation-based gaslighting is pervasive and severe, even among models with strong baseline performance. While prompt-level grounding constraints can partially mitigate this behavior, they do not reliably prevent **`hallucinated`** justifications or belief reversal. Our results indicate that current Vid-LLMs lack robust mechanisms for maintaining grounded spatiotemporal beliefs under adversarial conversational feedback. --- ### [When Text Hijacks Vision: Benchmarking and Mitigating Text Overlay-Induced Hallucination in Vision Language Models](https://arxiv.org/abs/2604.17375) - **arXiv ID:** `2604.17375` - **Date:** 04/2026 - **Code:** - - **Abstract:** Recent advances in Vision-Language Models (VLMs) have substantially enhanced their ability across multimodal video understanding benchmarks spanning temporal, action, object, and spatial understanding. However, we identify a critical yet overlooked issue: when embedded on-screen text contradicts the visual scene, existing VLMs systematically **`hallucinate`**, prioritizing overlay textual semantics over the actual visual content. We define this phenomenon as Text Overlay-Induced **`Hallucination`** (TOIH). In this work, we propose VisualTextTrap, the first comprehensive benchmark, including large-scale human-validated samples with specifically designed evaluation metrics. In particular, we construct VisualTextTrap from widely-used public datasets using a scalable hybrid pipeline of VLMs assisted text generation and rigorous manual verification. The benchmark features 6,057 samples annotated across 88 fine-grained attributes within four dimensions, with **`hallucination`** intensity quantified on a five-level scale (L1--L5) that reflects the semantic contradiction between overlay text and visual reality. Moreover, we propose Visual Text **`Hallucination`** Mitigation Mixture-of-Experts (VTHM-MoE), a novel Vision-Text Disentanglement framework that employs a dual-encoder architecture. Concretely, four dimension-specialized expert modules spanning Temporal, Action, Object, and Spatial reasoning are first pre-trained to identify and leverage cross-modal discrepancies between textual semantics and actual video content. We develop an Adaptive Token Routing Strategy to enable dynamic expert allocation, conferring robust resistance to TOIH while preserving performance on uncontaminated videos. Extensive experiments conducted on our VisualTextTrap benchmark verify the effectiveness of VTHM-MoE, outperforming state-of-the-art counterparts with diverse video question answering tasks. --- ### [Lyra 2.0: Explorable Generative 3D Worlds](https://arxiv.org/abs/2604.13036) - **arXiv ID:** `2604.13036` - **Date:** 04/2026 - **Code:** - - **Abstract:** Recent advances in video generation enable a new paradigm for 3D scene creation: generating camera-controlled videos that simulate scene walkthroughs, then lifting them to 3D via feed-forward reconstruction techniques. This generative reconstruction approach combines the visual fidelity and creative capacity of video models with 3D outputs ready for real-time rendering and simulation. Scaling to large, complex environments requires 3D-consistent video generation over long camera trajectories with large viewpoint changes and location revisits, a setting where current video models degrade quickly. Existing methods for long-horizon generation are fundamentally limited by two forms of degradation: spatial forgetting and temporal drifting. As exploration proceeds, previously observed regions fall outside the model's temporal context, forcing the model to **`hallucinate`** structures when revisited. Meanwhile, autoregressive generation accumulates small synthesis errors over time, gradually distorting scene appearance and geometry. We present Lyra 2.0, a framework for generating persistent, explorable 3D worlds at scale. To address spatial forgetting, we maintain per-frame 3D geometry and use it solely for information routing -- retrieving relevant past frames and establishing dense correspondences with the target viewpoints -- while relying on the generative prior for appearance synthesis. To address temporal drifting, we train with self-augmented histories that expose the model to its own degraded outputs, teaching it to correct drift rather than propagate it. Together, these enable substantially longer and 3D-consistent video trajectories, which we leverage to fine-tune feed-forward reconstruction models that reliably recover high-quality 3D scenes. --- ### [Relaxing Anchor-Frame Dominance for Mitigating Hallucinations in Video Large Language Models](https://arxiv.org/abs/2604.12582) - **arXiv ID:** `2604.12582` - **Date:** 04/2026 - **Code:** - - **Abstract:** Recent Video Large Language Models (Video-LLMs) have demonstrated strong capability in video understanding, yet they still suffer from **`hallucinations`**. Existing mitigation methods typically rely on training, input modification, auxiliary guidance, or additional decoding procedures, while largely overlooking a more fundamental challenge. During generation, Video-LLMs tend to over-rely on a limited portion of temporal evidence, leading to temporally imbalanced evidence aggregation across the video. To address this issue, we investigate a decoder-side phenomenon in which the model exhibits a temporally imbalanced concentration pattern. We term the frame with the highest aggregated frame-level attention mass the anchor frame. We find that this bias is largely independent of the input video and instead appears to reflect a persistent, model-specific structural or positional bias, whose over-dominance is closely associated with **`hallucination`**-prone generation. Motivated by this insight, we propose Decoder-side Temporal Rebalancing (DTR), a training-free, layer-selective inference method that rebalances temporal evidence allocation in middle-to-late decoder layers without altering visual encoding or requiring auxiliary models. DTR adaptively calibrates decoder-side visual attention to alleviate temporally imbalanced concentration and encourage under-attended frames to contribute more effectively to response generation. In this way, DTR guides the decoder to ground its outputs in temporally broader and more balanced video evidence. Extensive experiments on **`hallucination`** and video understanding benchmarks show that DTR consistently improves **`hallucination`** robustness across diverse Video-LLM families, while preserving competitive video understanding performance and high inference efficiency. --- ### [ArtifactWorld: Scaling 3D Gaussian Splatting Artifact Restoration via Video Generation Models](https://arxiv.org/abs/2604.12251) - **arXiv ID:** `2604.12251` - **Date:** 04/2026 - **Code:** - - **Abstract:** 3D Gaussian Splatting (3DGS) delivers high-fidelity real-time rendering but suffers from geometric and photometric degradations under sparse-view constraints. Current generative restoration approaches are often limited by insufficient temporal coherence, a lack of explicit spatial constraints, and a lack of large-scale training data, resulting in multi-view inconsistencies, erroneous geometric **`hallucinations`**, and limited generalization to diverse real-world artifact distributions. In this paper, we present ArtifactWorld, a framework that resolves 3DGS artifact repair through systematic data expansion and a homogeneous dual-model paradigm. To address the data bottleneck, we establish a fine-grained phenomenological taxonomy of 3DGS artifacts and construct a comprehensive training set of 107.5K diverse paired video clips to enhance model robustness. Architecturally, we unify the restoration process within a video diffusion backbone, utilizing an isomorphic predictor to localize structural defects via an artifact heatmap. This heatmap then guides the restoration through an Artifact-Aware Triplet Fusion mechanism, enabling precise, intensity-guided spatio-temporal repair within native self-attention. Extensive experiments demonstrate that ArtifactWorld achieves state-of-the-art performance in sparse novel view synthesis and robust 3D reconstruction. Code and dataset will be made public. --- ### [STEAR: Layer-Aware Spatiotemporal Evidence Intervention for Hallucination Mitigation in Video Large Language Models](https://arxiv.org/abs/2604.03045) - **arXiv ID:** `2604.03045` - **Date:** 04/2026 - **Code:** - - **Abstract:** Video Large Language Models (Video-LLMs) remain prone to spatiotemporal **`hallucinations`**, often generating visually unsupported details or incorrect temporal relations. Existing mitigation methods typically treat **`hallucination`** as a uniform decoding failure, applying globally shared correction rules. We instead observe that decoder layers contribute differently to visual grounding and later linguistic composition, indicating that intervention must be layer-aware. Based on this insight, we propose STEAR, a layer-aware spatiotemporal evidence intervention framework. STEAR identifies high-risk decoding steps and selects token-conditioned visual evidence from grounding-sensitive middle layers. It uses this shared evidence for two coupled purposes: restoring missing local grounding in middle layers, and constructing temporally perturbed patch-level counterfactuals to falsify inconsistent reasoning during late-layer decoding. Consequently, STEAR mitigates both spatial and temporal **`hallucinations`** within an efficient single-encode inference framework. Experiments across representative Video-LLM backbones and challenging benchmarks demonstrate that STEAR consistently reduces **`hallucinations`** while improving faithfulness, temporal consistency, and robustness. Our results confirm that reliable video decoding relies on intervening on precise evidence at the right layer, rather than enforcing a global penalty. The code is provided in the Supplementary Material. --- ### [Reinforcing Consistency in Video MLLMs with Structured Rewards](https://arxiv.org/abs/2604.01460) - **arXiv ID:** `2604.01460` - **Date:** 04/2026 - **Code:** - - **Abstract:** Multimodal large language models (MLLMs) have achieved remarkable progress in video understanding. However, seemingly plausible outputs often suffer from poor visual and temporal grounding: a model may fabricate object existence, assign incorrect attributes, or collapse repeated events while still producing a globally reasonable caption or answer. We study this failure mode through a compositional consistency audit that decomposes a caption into supporting factual and temporal claims, investigating whether a correct high-level prediction is actually backed by valid lower-level evidence. Our top-down audit reveals that even correct root relational claims often lack reliable attribute and existence support. This indicates that standard sentence-level supervision is a weak proxy for faithful video understanding. Furthermore, when turning to reinforcement learning (RL) for better alignment, standard sentence-level rewards often prove too coarse to accurately localize specific grounding failures. To address this, we replace generic sentence-level rewards with a structured reward built from factual and temporal units. Our training objective integrates three complementary components: (1) an instance-aware scene-graph reward for factual objects, attributes, and relations; (2) a temporal reward for event ordering and repetition; and (3) a video-grounded VQA reward for hierarchical self-verification. Across temporal, general video understanding, and **`hallucination`**-oriented benchmarks, this objective yields consistent gains on open-source backbones. These results suggest that structured reward shaping is a practical route to more faithful video understanding. --- ## Update — 2026-04-01 **13 new paper(s) found.** ### [VideoTIR: Accurate Understanding for Long Videos with Efficient Tool-Integrated Reasoning](https://arxiv.org/abs/2603.25021) - **arXiv ID:** `2603.25021` - **Date:** 03/2026 - **Code:** - - **Abstract:** Existing Multimodal Large Language Models (MLLMs) often suffer from **`hallucinations`** in long video understanding (LVU), primarily due to the imbalance between textual and visual tokens. Observing that MLLMs handle short visual inputs well, recent LVU works alleviate **`hallucinations`** by automatically parsing the vast visual data into manageable segments that can be effectively processed by MLLMs. SFT-based tool-calling methods can serve this purpose, but they typically require vast amounts of fine-grained, high-quality data and suffer from constrained tool-calling trajectories. We propose a novel VideoTIR that leverages Reinforcement Learning (RL) to encourage proper usage of comprehensive multi-level toolkits for efficient long video understanding. VideoTIR explores both Zero-RL and SFT cold-starting to enable MLLMs to retrieve and focus on meaningful video segments/images/regions, enhancing long video understanding both accurately and efficiently. To reduce redundant tool-calling, we propose Toolkit Action Grouped Policy Optimization (TAGPO), which enhances the efficiency of the calling process through stepwise reward assignment and reuse of failed rollouts. Additionally, we develop a sandbox-based trajectory synthesis framework to generate high-quality trajectories data. Extensive experiments on three long-video QA benchmarks demonstrate the effectiveness and efficiency of our method. --- ### [GameplayQA: A Benchmarking Framework for Decision-Dense POV-Synced Multi-Video Understanding of 3D Virtual Agents](https://arxiv.org/abs/2603.24329) - **arXiv ID:** `2603.24329` - **Date:** 03/2026 - **Code:** - - **Abstract:** Multimodal LLMs are increasingly deployed as perceptual backbones for autonomous agents in 3D environments, from robotics to virtual worlds. These applications require agents to perceive rapid state changes, attribute actions to the correct entities, and reason about concurrent multi-agent behaviors from a first-person perspective, capabilities that existing benchmarks do not adequately evaluate. We introduce GameplayQA, a framework for evaluating agentic-centric perception and reasoning through video understanding. Specifically, we densely annotate multiplayer 3D gameplay videos at 1.22 labels/second, with time-synced, concurrent captions of states, actions, and events structured around a triadic system of Self, Other Agents, and the World, a natural decomposition for multi-agent environments. From these annotations, we refined 2.4K diagnostic QA pairs organized into three levels of cognitive complexity, accompanied by a structured distractor taxonomy that enables fine-grained analysis of where models **`hallucinate`**. Evaluation of frontier MLLMs reveals a substantial gap from human performance, with common failures in temporal and cross-video grounding, agent-role attribution, and handling the decision density of the game. We hope GameplayQA stimulates future research at the intersection of embodied AI, agentic perception, and world modeling. --- ### [PAM: A Pose-Appearance-Motion Engine for Sim-to-Real HOI Video Generation](https://arxiv.org/abs/2603.22193) - **arXiv ID:** `2603.22193` - **Date:** 03/2026 - **Code:** [Code](https://github.com/GasaiYU/PAM) - **Abstract:** Hand-object interaction (HOI) reconstruction and synthesis are becoming central to embodied AI and AR/VR. Yet, despite rapid progress, existing HOI generation research remains fragmented across three disjoint tracks: (1) pose-only synthesis that predicts MANO trajectories without producing pixels; (2) single-image HOI generation that **`hallucinates`** appearance from masks or 2D cues but lacks dynamics; and (3) video generation methods that require both the entire pose sequence and the ground-truth first frame as inputs, preventing true sim-to-real deployment. Inspired by the philosophy of Joo et al. (2018), we think that HOI generation requires a unified engine that brings together pose, appearance, and motion within one coherent framework. Thus we introduce PAM: a Pose-Appearance-Motion Engine for controllable HOI video generation. The performance of our engine is validated by: (1) On DexYCB, we obtain an FVD of 29.13 (vs. 38.83 for InterDyn), and MPJPE of 19.37 mm (vs. 30.05 mm for CosHand), while generating higher-resolution 480x720 videos compared to 256x256 and 256x384 baselines. (2) On OAKINK2, our full multi-condition model improves FVD from 68.76 to 46.31. (3) An ablation over input conditions on DexYCB shows that combining depth, segmentation, and keypoints consistently yields the best results. (4) For a downstream hand pose estimation task using SimpleHand, augmenting training with 3,400 synthetic videos (207k frames) allows a model trained on only 50% of the real data plus our synthetic data to match the 100% real baseline. --- ### [MME-CoF-Pro: Evaluating Reasoning Coherence in Video Generative Models with Text and Visual Hints](https://arxiv.org/abs/2603.20194) - **arXiv ID:** `2603.20194` - **Date:** 03/2026 - **Code:** - - **Abstract:** Video generative models show emerging reasoning behaviors. It is essential to ensure that generated events remain causally consistent across frames for reliable deployment, a property we define as reasoning coherence. To bridge the gap in literature for missing reasoning coherence evaluation, we propose MME-CoF-Pro, a comprehensive video reasoning benchmark to assess reasoning coherence in video models. Specifically, MME-CoF-Pro contains 303 samples across 16 categories, ranging from visual logical to scientific reasoning. It introduces Reasoning Score as evaluation metric for assessing process-level necessary intermediate reasoning steps, and includes three evaluation settings, (a) no hint (b) text hint and (c) visual hint, enabling a controlled investigation into the underlying mechanisms of reasoning hint guidance. Evaluation results in 7 open and closed-source video models reveals insights including: (1) Video generative models exhibit weak reasoning coherence, decoupled from generation quality. (2) Text hints boost apparent correctness but often cause inconsistency and **`hallucinated`** reasoning (3) Visual hints benefit structured perceptual tasks but struggle with fine-grained perception. Website: https://video-reasoning-coherence.github.io/ --- ### [HopChain: Multi-Hop Data Synthesis for Generalizable Vision-Language Reasoning](https://arxiv.org/abs/2603.17024) - **arXiv ID:** `2603.17024` - **Date:** 03/2026 - **Code:** - - **Abstract:** Vision-language models (VLMs) show strong multimodal capabilities but still struggle with fine-grained vision-language reasoning. We find that long chain-of-thought (CoT) reasoning exposes diverse failure modes, including perception, reasoning, knowledge, and **`hallucination`** errors, which can compound across intermediate steps. However, most existing vision-language data used for reinforcement learning with verifiable rewards (RLVR) does not involve complex reasoning chains that rely on visual evidence throughout, leaving these weaknesses largely unexposed. We therefore propose HopChain, a scalable framework for synthesizing multi-hop vision-language reasoning data for RLVR training of VLMs. Each synthesized multi-hop query forms a logically dependent chain of instance-grounded hops, where earlier hops establish the instances, sets, or conditions needed for later hops, while the final answer remains a specific, unambiguous number suitable for verifiable rewards. We train Qwen3.5-35B-A3B and Qwen3.5-397B-A17B under two RLVR settings: the original data alone, and the original data plus HopChain's multi-hop data, and compare them across 24 benchmarks spanning STEM and Puzzle, General VQA, Text Recognition and Document Understanding, and Video Understanding. Although this multi-hop data is not synthesized for any specific benchmark, it improves 20 of 24 benchmarks on both models, indicating broad and generalizable gains. Consistently, replacing full chained queries with half-multi-hop or single-hop variants reduces the average score across five representative benchmarks from 70.4 to 66.7 and 64.3, respectively. Notably, multi-hop gains peak in long-CoT vision-language reasoning, exceeding 50 points in the ultra-long-CoT regime. These experiments establish HopChain as an effective, scalable framework for synthesizing multi-hop data that improves generalizable vision-language reasoning. --- ### [When Thinking Hurts: Mitigating Visual Forgetting in Video Reasoning via Frame Repetition](https://arxiv.org/abs/2603.16256) - **arXiv ID:** `2603.16256` - **Date:** 03/2026 - **Code:** - - **Abstract:** Recently, Multimodal Large Language Models (MLLMs) have demonstrated significant potential in complex visual tasks through the integration of Chain-of-Thought (CoT) reasoning. However, in Video Question Answering, extended thinking processes do not consistently yield performance gains and may even lead to degradation due to ``visual anchor drifting'', where models increasingly rely on self-generated text, sidelining visual inputs and causing **`hallucinations`**. While existing mitigations typically introduce specific mechanisms for the model to re-attend to visual inputs during inference, these approaches often incur prohibitive training costs and suffer from poor generalizability across different architectures. To address this, we propose FrameRepeat, an automated enhancement framework which features a lightweight repeat scoring module that enables Video-LLMs to autonomously identify which frames should be reinforced. We introduce a novel training strategy, Add-One-In (AOI), that uses MLLM output probabilities to generate supervision signals representing repeat gain. This can be used to train a frame scoring network, which guides the frame repetition behavior. Experimental results across multiple models and datasets demonstrate that FrameRepeat is both effective and generalizable in strengthening important visual cues during the reasoning process. --- ### [Clue Matters: Leveraging Latent Visual Clues to Empower Video Reasoning](https://arxiv.org/abs/2603.15008) - **arXiv ID:** `2603.15008` - **Date:** 03/2026 - **Code:** - - **Abstract:** Multi-modal Large Language Models (MLLMs) have significantly advanced video reasoning, yet Video Question Answering (VideoQA) remains challenging due to its demand for temporal causal reasoning and evidence-grounded answer generation. Prevailing end-to-end MLLM frameworks lack explicit structured reasoning between visual perception and answer derivation, causing severe **`hallucinations`** and poor interpretability. Existing methods also fail to address three core gaps: faithful visual clue extraction, utility-aware clue filtering, and end-to-end clue-answer alignment. Inspired by hierarchical human visual cognition, we propose ClueNet, a clue-aware video reasoning framework with a two-stage supervised fine-tuning paradigm without extensive base model modifications. Decoupled supervision aligns clue extraction and chain-based reasoning, while inference supervision with an adaptive clue filter refines high-order reasoning, alongside lightweight modules for efficient inference. Experiments on NExT-QA, STAR, and MVBench show that ClueNet outperforms state-of-the-art methods by $\ge$ 1.1%, with superior generalization, **`hallucination`** mitigation, inference efficiency, and cross-backbone compatibility. This work bridges the perception-to-generation gap in MLLM video understanding, providing an interpretable, faithful reasoning paradigm for high-stakes VideoQA applications. --- ### [The Pulse of Motion: Measuring Physical Frame Rate from Visual Dynamics](https://arxiv.org/abs/2603.14375) - **arXiv ID:** `2603.14375` - **Date:** 03/2026 - **Code:** - - **Abstract:** While recent generative video models have achieved remarkable visual realism and are being explored as world models, true physical simulation requires mastering both space and time. Current models can produce visually smooth kinematics, yet they lack a reliable internal motion pulse to ground these motions in a consistent, real-world time scale. This temporal ambiguity stems from the common practice of indiscriminately training on videos with vastly different real-world speeds, forcing them into standardized frame rates. This leads to what we term chronometric **`hallucination`**: generated sequences exhibit ambiguous, unstable, and uncontrollable physical motion speeds. To address this, we propose Visual Chronometer, a predictor that recovers the Physical Frames Per Second (PhyFPS) directly from the visual dynamics of an input video. Trained via controlled temporal resampling, our method estimates the true temporal scale implied by the motion itself, bypassing unreliable metadata. To systematically quantify this issue, we establish two benchmarks, PhyFPS-Bench-Real and PhyFPS-Bench-Gen. Our evaluations reveal a harsh reality: state-of-the-art video generators suffer from severe PhyFPS misalignment and temporal instability. Finally, we demonstrate that applying PhyFPS corrections significantly improves the human-perceived naturalness of AI-generated videos. Our project page is https://xiangbogaobarry.github.io/Visual_Chronometer/. --- ### [Event-Driven Video Generation](https://arxiv.org/abs/2603.13402) - **arXiv ID:** `2603.13402` - **Date:** 03/2026 - **Code:** - - **Abstract:** State-of-the-art text-to-video models often look realistic frame-by-frame yet fail on simple interactions: motion starts before contact, actions are not realized, objects drift after placement, and support relations break. We argue this stems from frame-first denoising, which updates latent state everywhere at every step without an explicit notion of when and where an interaction is active. We introduce Event-Driven Video Generation (EVD), a minimal DiT-compatible framework that makes sampling event-grounded: a lightweight event head predicts token-aligned event activity, event-grounded losses couple activity to state change during training, and event-gated sampling (with hysteresis and early-step scheduling) suppresses spurious updates while concentrating updates during interactions. On EVD-Bench, EVD consistently improves human preference and VBench dynamics, substantially reducing failure modes in state persistence, spatial accuracy, support relations, and contact stability without sacrificing appearance. These results indicate that explicit event grounding is a practical abstraction for reducing interaction **`hallucinations`** in video generation. --- ### [Unleashing Video Language Models for Fine-grained HRCT Report Generation](https://arxiv.org/abs/2603.12469) - **arXiv ID:** `2603.12469` - **Date:** 03/2026 - **Code:** - - **Abstract:** Generating precise diagnostic reports from High-Resolution Computed Tomography (HRCT) is critical for clinical workflow, yet it remains a formidable challenge due to the high pathological diversity and spatial sparsity within 3D volumes. While Video Language Models (VideoLMs) have demonstrated remarkable spatio-temporal reasoning in general domains, their adaptability to domain-specific, high-volume medical interpretation remains underexplored. In this work, we present AbSteering, an abnormality-centric framework that steers VideoLMs toward precise HRCT report generation. Specifically, AbSteering introduces: (i) an abnormality-centric Chain-of-Thought scheme that enforces abnormality reasoning, and (ii) a Direct Preference Optimization objective that utilizes clinically confusable abnormalities as hard negatives to enhance fine-grained discrimination. Our results demonstrate that general-purpose VideoLMs possess strong transferability to high-volume medical imaging when guided by this paradigm. Notably, AbSteering outperforms state-of-the-art domain-specific CT foundation models, which are pretrained with large-scale CTs, achieving superior detection sensitivity while simultaneously mitigating **`hallucinations`**. --- ### [Controllable Egocentric Video Generation via Occlusion-Aware Sparse 3D Hand Joints](https://arxiv.org/abs/2603.11755) - **arXiv ID:** `2603.11755` - **Date:** 03/2026 - **Code:** - - **Abstract:** Motion-controllable video generation is crucial for egocentric applications in virtual reality and embodied AI. However, existing methods often struggle to achieve 3D-consistent fine-grained hand articulation. By adopting on 2D trajectories or implicit poses, they collapse 3D geometry into spatially ambiguous signals or over rely on human-centric priors. Under severe egocentric occlusions, this causes motion inconsistencies and **`hallucinated`** artifacts, as well as preventing cross-embodiment generalization to robotic hands. To address these limitations, we propose a novel framework that generates egocentric videos from a single reference frame, leveraging sparse 3D hand joints as embodiment-agnostic control signals with clear semantic and geometric structures. We introduce an efficient control module that resolves occlusion ambiguities while fully preserving 3D information. Specifically, it extracts occlusion-aware features from the source reference frame by penalizing unreliable visual signals from hidden joints, and employs a 3D-based weighting mechanism to robustly handle dynamically occluded target joints during motion propagation. Concurrently, the module directly injects 3D geometric embeddings into the latent space to strictly enforce structural consistency. To facilitate robust training and evaluation, we develop an automated annotation pipeline that yields over one million high-quality egocentric video clips paired with precise hand trajectories. Additionally, we register humanoid kinematic and camera data to construct a cross-embodiment benchmark. Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art baselines, generating high-fidelity egocentric videos with realistic interactions and exhibiting exceptional cross-embodiment generalization to robotic hands. --- ### [INFACT: A Diagnostic Benchmark for Induced Faithfulness and Factuality Hallucinations in Video-LLMs](https://arxiv.org/abs/2603.11481) - **arXiv ID:** `2603.11481` - **Date:** 03/2026 - **Code:** - - **Abstract:** 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. --- ### [World2Act: Latent Action Post-Training via Skill-Compositional World Models](https://arxiv.org/abs/2603.10422) - **arXiv ID:** `2603.10422` - **Date:** 03/2026 - **Code:** - - **Abstract:** World Models (WMs) have emerged as a promising approach for post-training Vision-Language-Action (VLA) policies to improve robustness and generalization under environmental changes. However, most WM-based post-training methods rely on pixel-space supervision, making policies sensitive to pixel-level artifacts and **`hallucination`** from imperfect WM rollouts. We introduce World2Act, a post-training framework that aligns VLA actions directly with WM video-dynamics latents using a contrastive matching objective, reducing dependence on pixels. Post-training performance is tied to rollout quality, yet current WMs struggle with arbitrary-length video generation as they are mostly trained on fixed-length clips while robotic execution durations vary widely. To address this, we propose an automatic LLM-based skill-decomposition pipeline that segments high-level instructions into low-level prompts. Our pipeline produces RoboCasa-Skill and LIBERO-Skill, supporting skill-compositional WMs that remain temporally consistent across diverse task horizons. Empirically, applying World2Act to VLAs like GR00T-N1.6 and Cosmos Policy achieves state-of-the-art results on RoboCasa and LIBERO, and improves real-world performance by 6.7%, enhancing embodied agent generalization. --- **18 new paper(s) found.** ### [From Verbatim to Gist: Distilling Pyramidal Multimodal Memory via Semantic Information Bottleneck for Long-Horizon Video Agents](https://arxiv.org/abs/2603.01455) - **arXiv ID:** `2603.01455` - **Date:** 03/2026 - **Code:** [Code](https://github.com/EliSpectre/MM-Mem) - **Abstract:** While multimodal large language models have demonstrated impressive short-term reasoning, they struggle with long-horizon video understanding due to limited context windows and static memory mechanisms that fail to mirror human cognitive efficiency. Existing paradigms typically fall into two extremes: vision-centric methods that incur high latency and redundancy through dense visual accumulation, or text-centric approaches that suffer from detail loss and **`hallucination`** via aggressive captioning. To bridge this gap, we propose MM-Mem, a pyramidal multimodal memory architecture grounded in Fuzzy-Trace Theory. MM-Mem structures memory hierarchically into a Sensory Buffer, Episodic Stream, and Symbolic Schema, enabling the progressive distillation of fine-grained perceptual traces (verbatim) into high-level semantic schemas (gist). Furthermore, to govern the dynamic construction of memory, we derive a Semantic Information Bottleneck objective and introduce SIB-GRPO to optimize the trade-off between memory compression and task-relevant information retention. In inference, we design an entropy-driven top-down memory retrieval strategy, which first tries with the abstract Symbolic Schema and progressively "drills down" to the Sensory Buffer and Episodic Stream under high uncertainty. Extensive experiments across 4 benchmarks confirm the effectiveness of MM-Mem on both offline and streaming tasks, demonstrating robust generalization and validating the effectiveness of cognition-inspired memory organization. Code is available at https://github.com/EliSpectre/MM-Mem. --- ### [Think with Grounding: Curriculum Reinforced Reasoning with Video Grounding for Long Video Understanding](https://arxiv.org/abs/2602.18702) - **arXiv ID:** `2602.18702` - **Date:** 02/2026 - **Code:** - - **Abstract:** Long video understanding is challenging due to rich and complicated multimodal clues in long temporal range.Current methods adopt reasoning to improve the model's ability to analyze complex video clues in long videos via text-form reasoning.However,the existing literature suffers from the fact that the text-only reasoning under fixed video context may exacerbate **`hallucinations`** since detailed crucial clues are often ignored under limited video context length due to the temporal redundancy of long videos.To address this gap,we propose Video-TwG,a curriculum reinforced framework that employs a novel Think-with-Grounding paradigm,enabling video LLMs to actively decide when to perform on-demand grounding during interleaved text-video reasoning, selectively zooming into question-relevant clips only when necessary.Video-TwG can be trained end-to-end in a straightforward manner, without relying on complex auxiliary modules or heavily annotated reasoning tracesIn detail,we design a Two-stage Reinforced Curriculum Strategy, where the model first learns think-with-grounding behavior on a small short-video GQA dataset with grounding labels,and then scales to diverse general QA data with videos of diverse domains to encourage generalization. Further, to handle complex think-with-grounding reasoning for various kinds of data,we propose TwG-GRPO algorithm which features the fine-grained grounding reward, self-confirmed pseudo reward and accuracy-gated mechanism.Finally,we propose to construct a new TwG-51K dataset that facilitates training. Experiments on Video-MME, LongVideoBench, and MLVU show that Video-TwG consistently outperforms strong LVU baselines.Further ablation validates the necessity of our Two-stage Reinforced Curriculum Strategy and shows our TwG-GRPO better leverages diverse unlabeled data to improve grounding quality and reduce redundant groundings without sacrificing QA performance. --- ### [KPM-Bench: A Kinematic Parsing Motion Benchmark for Fine-grained Motion-centric Video Understanding](https://arxiv.org/abs/2602.17768) - **arXiv ID:** `2602.17768` - **Date:** 02/2026 - **Code:** - - **Abstract:** Despite recent advancements, video captioning models still face significant limitations in accurately describing fine-grained motion details and suffer from severe **`hallucination`** issues. These challenges become particularly prominent when generating captions for motion-centric videos, where precise depiction of intricate movements and limb dynamics is crucial yet often neglected. To alleviate this gap, we introduce an automated annotation pipeline that integrates kinematic-based motion computation with linguistic parsing, enabling detailed decomposition and description of complex human motions. Based on this pipeline, we construct and release the Kinematic Parsing Motion Benchmark (KPM-Bench), a novel open-source dataset designed to facilitate fine-grained motion understanding. KPM-Bench consists of (i) fine-grained video-caption pairs that comprehensively illustrate limb-level dynamics in complex actions, (ii) diverse and challenging question-answer pairs focusing specifically on motion understanding, and (iii) a meticulously curated evaluation set specifically designed to assess **`hallucination`** phenomena associated with motion descriptions. Furthermore, to address **`hallucination`** issues systematically, we propose the linguistically grounded Motion Parsing and Extraction (MoPE) algorithm, capable of accurately extracting motion-specific attributes directly from textual captions. Leveraging MoPE, we introduce a precise **`hallucination`** evaluation metric that functions independently of large-scale vision-language or language-only models. By integrating MoPE into the GRPO post-training framework, we effectively mitigate **`hallucination`** problems, significantly improving the reliability of motion-centric video captioning models. --- ### [GraphThinker: Reinforcing Video Reasoning with Event Graph Thinking](https://arxiv.org/abs/2602.17555) - **arXiv ID:** `2602.17555` - **Date:** 02/2026 - **Code:** - - **Abstract:** Video reasoning requires understanding the causal relationships between events in a video. However, such relationships are often implicit and costly to annotate manually. While existing multimodal large language models (MLLMs) often infer event relations through dense captions or video summaries for video reasoning, such modeling still lacks causal understanding. Without explicit causal structure modeling within and across video events, these models suffer from **`hallucinations`** during the video reasoning. In this work, we propose GraphThinker, a reinforcement finetuning-based method that constructs structural event-level scene graphs and enhances visual grounding to jointly reduce **`hallucinations`** in video reasoning. Specifically, we first employ an MLLM to construct an event-based video scene graph (EVSG) that explicitly models both intra- and inter-event relations, and incorporate these formed scene graphs into the MLLM as an intermediate thinking process. We also introduce a visual attention reward during reinforcement finetuning, which strengthens video grounding and further mitigates **`hallucinations`**. We evaluate GraphThinker on two datasets, RexTime and VidHalluc, where it shows superior ability to capture object and event relations with more precise event localization, reducing **`hallucinations`** in video reasoning compared to prior methods. --- ### [Towards Universal Video MLLMs with Attribute-Structured and Quality-Verified Instructions](https://arxiv.org/abs/2602.13013) - **arXiv ID:** `2602.13013` - **Date:** 02/2026 - **Code:** - - **Abstract:** Universal video understanding requires modeling fine-grained visual and audio information over time in diverse real-world scenarios. However, the performance of existing models is primarily constrained by video-instruction data that represents complex audiovisual content as single, incomplete descriptions, lacking fine-grained organization and reliable annotation. To address this, we introduce: (i) ASID-1M, an open-source collection of one million structured, fine-grained audiovisual instruction annotations with single- and multi-attribute supervision; (ii) ASID-Verify, a scalable data curation pipeline for annotation, with automatic verification and refinement that enforces semantic and temporal consistency between descriptions and the corresponding audiovisual content; and (iii) ASID-Captioner, a video understanding model trained via Supervised Fine-Tuning (SFT) on the ASID-1M. Experiments across seven benchmarks covering audiovisual captioning, attribute-wise captioning, caption-based QA, and caption-based temporal grounding show that ASID-Captioner improves fine-grained caption quality while reducing **`hallucinations`** and improving instruction following. It achieves state-of-the-art performance among open-source models and is competitive with Gemini-3-Pro. --- ### [STVG-R1: Incentivizing Instance-Level Reasoning and Grounding in Videos via Reinforcement Learning](https://arxiv.org/abs/2602.11730) - **arXiv ID:** `2602.11730` - **Date:** 02/2026 - **Code:** - - **Abstract:** In vision-language models (VLMs), misalignment between textual descriptions and visual coordinates often induces **`hallucinations`**. This issue becomes particularly severe in dense prediction tasks such as spatial-temporal video grounding (STVG). Prior approaches typically focus on enhancing visual-textual alignment or attaching auxiliary decoders. However, these strategies inevitably introduce additional trainable modules, leading to significant annotation costs and computational overhead. In this work, we propose a novel visual prompting paradigm that avoids the difficult problem of aligning coordinates across modalities. Specifically, we reformulate per-frame coordinate prediction as a compact instance-level identification problem by assigning each object a unique, temporally consistent ID. These IDs are embedded into the video as visual prompts, providing explicit and interpretable inputs to the VLMs. Furthermore, we introduce STVG-R1, the first reinforcement learning framework for STVG, which employs a task-driven reward to jointly optimize temporal accuracy, spatial consistency, and structural format regularization. Extensive experiments on six benchmarks demonstrate the effectiveness of our approach. STVG-R1 surpasses the baseline Qwen2.5-VL-7B by a remarkable margin of 20.9% on m_IoU on the HCSTVG-v2 benchmark, establishing a new state of the art (SOTA). Surprisingly, STVG-R1 also exhibits strong zero-shot generalization to multi-object referring video object segmentation tasks, achieving a SOTA 47.3% J&F on MeViS. --- ### [Reliable and Responsible Foundation Models: A Comprehensive Survey](https://arxiv.org/abs/2602.08145) - **arXiv ID:** `2602.08145` - **Date:** 02/2026 - **Code:** - - **Abstract:** Foundation models, including Large Language Models (LLMs), Multimodal Large Language Models (MLLMs), Image Generative Models (i.e, Text-to-Image Models and Image-Editing Models), and Video Generative Models, have become essential tools with broad applications across various domains such as law, medicine, education, finance, science, and beyond. As these models see increasing real-world deployment, ensuring their reliability and responsibility has become critical for academia, industry, and government. This survey addresses the reliable and responsible development of foundation models. We explore critical issues, including bias and fairness, security and privacy, uncertainty, explainability, and distribution shift. Our research also covers model limitations, such as **`hallucinations`**, as well as methods like alignment and Artificial Intelligence-Generated Content (AIGC) detection. For each area, we review the current state of the field and outline concrete future research directions. Additionally, we discuss the intersections between these areas, highlighting their connections and shared challenges. We hope our survey fosters the development of foundation models that are not only powerful but also ethical, trustworthy, reliable, and socially responsible. --- ### [VideoTemp-o3: Harmonizing Temporal Grounding and Video Understanding in Agentic Thinking-with-Videos](https://arxiv.org/abs/2602.07801) - **arXiv ID:** `2602.07801` - **Date:** 02/2026 - **Code:** - - **Abstract:** In long-video understanding, conventional uniform frame sampling often fails to capture key visual evidence, leading to degraded performance and increased **`hallucinations`**. To address this, recent agentic thinking-with-videos paradigms have emerged, adopting a localize-clip-answer pipeline in which the model actively identifies relevant video segments, performs dense sampling within those clips, and then produces answers. However, existing methods remain inefficient, suffer from weak localization, and adhere to rigid workflows. To solve these issues, we propose VideoTemp-o3, a unified agentic thinking-with-videos framework that jointly models video grounding and question answering. VideoTemp-o3 exhibits strong localization capability, supports on-demand clipping, and can refine inaccurate localizations. Specifically, in the supervised fine-tuning stage, we design a unified masking mechanism that encourages exploration while preventing noise. For reinforcement learning, we introduce dedicated rewards to mitigate reward hacking. Besides, from the data perspective, we develop an effective pipeline to construct high-quality long video grounded QA data, along with a corresponding benchmark for systematic evaluation across various video durations. Experimental results demonstrate that our method achieves remarkable performance on both long video understanding and grounding. --- ### [Process-of-Thought Reasoning for Videos](https://arxiv.org/abs/2602.07689) - **arXiv ID:** `2602.07689` - **Date:** 02/2026 - **Code:** - - **Abstract:** Video understanding requires not only recognizing visual content but also performing temporally grounded, multi-step reasoning over long and noisy observations. We propose Process-of-Thought (PoT) Reasoning for Videos, a framework that makes the reasoning process explicit by structuring video inference into a sequence of lightweight, verifiable steps. PoT interleaves (i) temporal evidence selection, (ii) step-wise state updates, and (iii) constrained answer synthesis, enabling the model to progressively refine hypotheses while maintaining traceability to video evidence. The framework is designed to be model-agnostic and can be plugged into existing vision-language backbones, supporting both closed-book reasoning and evidence-augmented reasoning with external tools. We further introduce a unified representation for PoT traces that aligns intermediate decisions with temporal segments, which improves robustness to distractors and reduces **`hallucinated`** explanations. Extensive experiments on standard video reasoning tasks demonstrate that PoT consistently improves factual correctness and temporal grounding, while providing interpretable reasoning traces for diagnosis and downstream use. --- ### [MACD: Model-Aware Contrastive Decoding via Counterfactual Data](https://arxiv.org/abs/2602.01740) - **arXiv ID:** `2602.01740` - **Date:** 02/2026 - **Code:** - - **Abstract:** Video language models (Video-LLMs) are prone to **`hallucinations`**, often generating plausible but ungrounded content when visual evidence is weak, ambiguous, or biased. Existing decoding methods, such as contrastive decoding (CD), rely on random perturbations to construct contrastive data for mitigating **`hallucination`** patterns. However, such a way is hard to control the visual cues that drive **`hallucination`** or well align with model weaknesses. We propose Model-aware Counterfactual Data based Contrastive Decoding (MACD), a new inference strategy that combines model-guided counterfactual construction with decoding. Our approach uses the Video-LLM's own feedback to identify object regions most responsible for **`hallucination`**, generating targeted counterfactual inputs at the object level rather than arbitrary frame or temporal modifications. These model-aware counterfactual data is then integrated into CD to enforce evidence-grounded token selection during decoding. Experiments on EventHallusion, MVBench, Perception-test and Video-MME show that MACD consistently reduces **`hallucination`** while maintaining or improving task accuracy across diverse Video-LLMs, including Qwen and InternVL families. The method is especially effective in challenging scenarios involving small, occluded, or co-occurring objects. Our code and data will be publicly released. --- ### [Learning to Decode Against Compositional Hallucination in Video Multimodal Large Language Models](https://arxiv.org/abs/2602.00559) - **arXiv ID:** `2602.00559` - **Date:** 01/2026 - **Code:** [Code](https://github.com/BMRETURN/OmniVCHall) - **Abstract:** Current research on video **`hallucination`** mitigation primarily focuses on isolated error types, leaving compositional **`hallucinations`**, arising from incorrect reasoning over multiple interacting spatial and temporal factors largely underexplored. We introduce OmniVCHall, a benchmark designed to systematically evaluate both isolated and compositional **`hallucinations`** in video multimodal large language models (VLLMs). OmniVCHall spans diverse video domains, introduces a novel camera-based **`hallucination`** type, and defines a fine-grained taxonomy, together with adversarial answer options (e.g., "All are correct" and "None of the above") to prevent shortcut reasoning. The evaluations of 39 representative VLLMs reveal that even advanced models (e.g., Qwen3-VL and GPT-5) exhibit substantial performance degradation. We propose TriCD, a contrastive decoding framework with a triple-pathway calibration mechanism. An adaptive perturbation controller dynamically selects distracting operations to construct negative video variants, while a saliency-guided enhancement module adaptively reinforces grounded token-wise visual evidences. These components are optimized via reinforcement learning to encourage precise decision-making under compositional **`hallucination`** settings. Experimental results show that TriCD consistently improves performance across two representative backbones, achieving an average accuracy improvement of over 10%. The data and code can be find at https://github.com/BMRETURN/OmniVCHall. --- ### [Mitigating Hallucinations in Video Large Language Models via Spatiotemporal-Semantic Contrastive Decoding](https://arxiv.org/abs/2601.22574) - **arXiv ID:** `2601.22574` - **Date:** 01/2026 - **Code:** - - **Abstract:** Although Video Large Language Models perform remarkably well across tasks such as video understanding, question answering, and reasoning, they still suffer from the problem of **`hallucination`**, which refers to generating outputs that are inconsistent with explicit video content or factual evidence. However, existing decoding methods for mitigating video **`hallucinations`**, while considering the spatiotemporal characteristics of videos, mostly rely on heuristic designs. As a result, they fail to precisely capture the root causes of **`hallucinations`** and their fine-grained temporal and semantic correlations, leading to limited robustness and generalization in complex scenarios. To more effectively mitigate video **`hallucinations`**, we propose a novel decoding strategy termed Spatiotemporal-Semantic Contrastive Decoding. This strategy constructs negative features by deliberately disrupting the spatiotemporal consistency and semantic associations of video features, and suppresses video **`hallucinations`** through contrastive decoding against the original video features during inference. Extensive experiments demonstrate that our method not only effectively mitigates the occurrence of **`hallucinations`**, but also preserves the general video understanding and reasoning capabilities of the model. --- ### [VideoHEDGE: Entropy-Based Hallucination Detection for Video-VLMs via Semantic Clustering and Spatiotemporal Perturbations](https://arxiv.org/abs/2601.08557) - **arXiv ID:** `2601.08557` - **Date:** 01/2026 - **Code:** [Code](https://github.com/Simula/HEDGE) - **Abstract:** **`Hallucinations`** in video-capable vision-language models (Video-VLMs) remain frequent and high-confidence, while existing uncertainty metrics often fail to align with correctness. We introduce VideoHEDGE, a modular framework for **`hallucination`** detection in video question answering that extends entropy-based reliability estimation from images to temporally structured inputs. Given a video-question pair, VideoHEDGE draws a baseline answer and multiple high-temperature generations from both clean clips and photometrically and spatiotemporally perturbed variants, then clusters the resulting textual outputs into semantic hypotheses using either Natural Language Inference (NLI)-based or embedding-based methods. Cluster-level probability masses yield three reliability scores: Semantic Entropy (SE), RadFlag, and Vision-Amplified Semantic Entropy (VASE). We evaluate VideoHEDGE on the SoccerChat benchmark using an LLM-as-a-judge to obtain binary **`hallucination`** labels. Across three 7B Video-VLMs (Qwen2-VL, Qwen2.5-VL, and a SoccerChat-finetuned model), VASE consistently achieves the highest ROC-AUC, especially at larger distortion budgets, while SE and RadFlag often operate near chance. We further show that embedding-based clustering matches NLI-based clustering in detection performance at substantially lower computational cost, and that domain fine-tuning reduces **`hallucination`** frequency but yields only modest improvements in calibration. The hedge-bench PyPI library enables reproducible and extensible benchmarking, with full code and experimental resources available at https://github.com/Simula/HEDGE#videohedge . --- ### [CASHEW: Stabilizing Multimodal Reasoning via Iterative Trajectory Aggregation](https://arxiv.org/abs/2601.08010) - **arXiv ID:** `2601.08010` - **Date:** 01/2026 - **Code:** - - **Abstract:** Vision-language models achieve strong performance across a wide range of multimodal understanding and reasoning tasks, yet their multi-step reasoning remains unstable. Repeated sampling over the same input often produces divergent reasoning trajectories and inconsistent final predictions. To address this, we introduce two complementary approaches inspired by test-time scaling: (1) CASHEW, an inference-time framework that stabilizes reasoning by iteratively aggregating multiple candidate trajectories into higher-quality reasoning traces, with explicit visual verification filtering **`hallucinated`** steps and grounding reasoning in visual evidence, and (2) CASHEW-RL, a learned variant that internalizes this aggregation behavior within a single model. CASHEW-RL is trained using Group Sequence Policy Optimization (GSPO) with a composite reward that encourages correct answers grounded in minimal yet sufficient visual evidence, while adaptively allocating reasoning effort based on task difficulty. This training objective enables robust self-aggregation at inference. Extensive experiments on 13 image understanding, video understanding, and video reasoning benchmarks show significant performance improvements, including gains of up to +23.6 percentage points on ScienceQA and +8.1 percentage points on EgoSchema. --- ### [Video Generation Models in Robotics -- Applications, Research Challenges, Future Directions](https://arxiv.org/abs/2601.07823) - **arXiv ID:** `2601.07823` - **Date:** 01/2026 - **Code:** - - **Abstract:** Video generation models have emerged as high-fidelity models of the physical world, capable of synthesizing high-quality videos capturing fine-grained interactions between agents and their environments conditioned on multi-modal user inputs. Their impressive capabilities address many of the long-standing challenges faced by physics-based simulators, driving broad adoption in many problem domains, e.g., robotics. For example, video models enable photorealistic, physically consistent deformable-body simulation without making prohibitive simplifying assumptions, which is a major bottleneck in physics-based simulation. Moreover, video models can serve as foundation world models that capture the dynamics of the world in a fine-grained and expressive way. They thus overcome the limited expressiveness of language-only abstractions in describing intricate physical interactions. In this survey, we provide a review of video models and their applications as embodied world models in robotics, encompassing cost-effective data generation and action prediction in imitation learning, dynamics and rewards modeling in reinforcement learning, visual planning, and policy evaluation. Further, we highlight important challenges hindering the trustworthy integration of video models in robotics, which include poor instruction following, **`hallucinations`** such as violations of physics, and unsafe content generation, in addition to fundamental limitations such as significant data curation, training, and inference costs. We present potential future directions to address these open research challenges to motivate research and ultimately facilitate broader applications, especially in safety-critical settings. --- ### [Video Evidence to Reasoning Efficient Video Understanding via Explicit Evidence Grounding](https://arxiv.org/abs/2601.07761) - **arXiv ID:** `2601.07761` - **Date:** 01/2026 - **Code:** - - **Abstract:** Large Vision-Language Models (LVLMs) face a fundamental dilemma in video reasoning: they are caught between the prohibitive computational costs of verbose reasoning and the **`hallucination`** risks of efficient, ungrounded approaches. To resolve this, we introduce the Chain of Evidence (CoE), a novel framework that architecturally decouples and co-optimizes perceptual grounding and reasoning efficiency. CoE incorporates two core innovations: (1) A lightweight Evidence Grounding Module (EGM) that acts as a query-guided filter, dynamically identifying and extracting a compact set of high-fidelity visual evidence; and (2) An Evidence-Anchoring Protocol optimized via Reinforcement Learning. Crucially, we design a composite reward mechanism that enforces process alignment, compelling the model to strictly reference identified temporal anchors during deduction, thereby mitigating **`hallucinations`**. To enable this, we construct CoE-Instruct, a large-scale dataset (164k samples) featuring a novel dual-annotation schema for separate perception and reasoning supervision. Extensive experiments on five benchmarks, including Video-MME, MVBench, and VSI-Bench, demonstrate that CoE-enhanced models establish a new state-of-the-art. They significantly outperform existing methods in accuracy, proving CoE to be a powerful and practical paradigm for reliable video understanding. --- ### [Plenoptic Video Generation](https://arxiv.org/abs/2601.05239) - **arXiv ID:** `2601.05239` - **Date:** 01/2026 - **Code:** - - **Abstract:** Camera-controlled generative video re-rendering methods, such as ReCamMaster, have achieved remarkable progress. However, despite their success in single-view setting, these works often struggle to maintain consistency across multi-view scenarios. Ensuring spatio-temporal coherence in **`hallucinated`** regions remains challenging due to the inherent stochasticity of generative models. To address it, we introduce PlenopticDreamer, a framework that synchronizes generative **`hallucinations`** to maintain spatio-temporal memory. The core idea is to train a multi-in-single-out video-conditioned model in an autoregressive manner, aided by a camera-guided video retrieval strategy that adaptively selects salient videos from previous generations as conditional inputs. In addition, Our training incorporates progressive context-scaling to improve convergence, self-conditioning to enhance robustness against long-range visual degradation caused by error accumulation, and a long-video conditioning mechanism to support extended video generation. Extensive experiments on the Basic and Agibot benchmarks demonstrate that PlenopticDreamer achieves state-of-the-art video re-rendering, delivering superior view synchronization, high-fidelity visuals, accurate camera control, and diverse view transformations (e.g., third-person to third-person, and head-view to gripper-view in robotic manipulation). Project page: https://research.nvidia.com/labs/dir/plenopticdreamer/ --- ### [CounterVid: Counterfactual Video Generation for Mitigating Action and Temporal Hallucinations in Video-Language Models](https://arxiv.org/abs/2601.04778) - **arXiv ID:** `2601.04778` - **Date:** 01/2026 - **Code:** - - **Abstract:** Video-language models (VLMs) achieve strong multimodal understanding but remain prone to **`hallucinations`**, especially when reasoning about actions and temporal order. Existing mitigation strategies, such as textual filtering or random video perturbations, often fail to address the root cause: over-reliance on language priors rather than fine-grained visual dynamics. We propose a scalable framework for counterfactual video generation that synthesizes videos differing only in actions or temporal structure while preserving scene context. Our pipeline combines multimodal LLMs for action proposal and editing guidance with diffusion-based image and video models to generate semantic hard negatives at scale. Using this framework, we build CounterVid, a synthetic dataset of ~26k preference pairs targeting action recognition and temporal reasoning. We further introduce MixDPO, a unified Direct Preference Optimization approach that jointly leverages textual and visual preferences. Fine-tuning Qwen2.5-VL with MixDPO yields consistent improvements, notably in temporal ordering, and transfers effectively to standard video **`hallucination`** benchmarks. Code and models will be made publicly available. ---