PyTorchKR Papers

최신 논문

총 418편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2611.06415v1cs.CV

Entropogram-guided Visual Realignment Decoding: Capturing the Rhythm of Multimodal Reasoning

Multimodal large language models (MLLMs) have demonstrated remarkable progress on cross-modal reasoning tasks, yet they continue to suffer from hallucinations in long-chain reasoning processes. In this work, we revisit hallucination in MLLMs from the perspective of reasoning generation dynamics. We argue that entropy is not merely a static measure of uncertainty but a structured signal that reflects how modality information is integrated throughout decoding. Through systematic analysis, we identify two characteristic entropy patterns associated with hallucination behaviors: (1) a generally high-confidence regime corresponds to premature convergence to inappropriate multimodal interactions; and (2) local abrupt entropy transitions are associated with reasoning instability and hallucination onset. Building on these observations, we propose an entropy-aware self-introspective framework for MLLM reasoning. Our method continuously tracks entropy dynamics during decoding and performs targeted visual re-alignment when suspicious hallucination states are detected, encouraging the model to re-examine salient visual evidence at critical reasoning stages. Extensive experiments across six multimodal reasoning benchmarks spanning nine distinct evaluation settings demonstrate that our approach delivers robust improvements in reasoning accuracy and hallucination robustness, while providing new insights into the generation dynamics underlying MLLM failures. The code is available at: https://github.com/jden4524/SelfRetrospectiveDecoding.

arXiv:2611.05992v1cs.CV

Clips2Story: Training-Free Video Storyboarding and Editing Using Multimodal Retrieval-Embedded Generation

Video editors often work with large collections of fragmented footage and must select and organize relevant clips into a coherent story. However, many existing video editing and summarization methods assume a temporally ordered source video and aim to preserve its original narrative and temporal structure. In this paper, we introduce Clips2Story, a training-free, LLM-based pipeline that enables users to create keyword-specified short videos from a pool of video clips. Given a pool of clips and a user-specified keyword, we first extract text-based multimodal descriptors for each clip and retrieve the clips most relevant to the keyword. We then provide the textual descriptors of these candidate clips to an LLM, which selects and arranges them into an editable storyboard with connective narrations. Finally, we assemble the selected clips and retrieve accompanying visuals for the narrations to compose the output video. To evaluate the proposed pipeline, we construct a synthetic evaluation dataset from 100 long-form videos across five genres by segmenting each video into short clips using a scene-change detection model and shuffling the resulting clips to remove their temporal order. Through objective evaluation and qualitative examples, we demonstrate that Clips2Story can effectively retrieve keyword-relevant clips and compose coherent storyboards based on user-specified keywords. Human evaluation further shows that Clips2Story outperforms a strong baseline in terms of narrative coherence, keyword alignment, and overall realism.

arXiv:2611.05893v1cs.CV

Chronos: Sequential Identity-Preserving Video Generation by Temporal Anchoring

Identity-preserving video generation aims to synthesize high-quality and visually realistic videos from textual and visual conditions while faithfully maintaining the subject's identity across diverse motions. Prior methods have extensively introduced reference-guided control signals and identity-aware conditioning to improve facial alignment from reference images. However, improving static identity alignment alone creates a fundamental tension between identity fidelity and temporal controllability, which frequently causes identity drift or motion suppression under large pose variations and multi-segment sequential actions. To address this, we propose Chronos, a framework that unifies single-segment facial identity preservation and multi-segment sequential action generation under a common temporal formulation. Specifically, Chronos integrates Semantic Prompt Anchoring to adaptively refine the user prompt, Visual Frame Anchoring to transform conditions into visually grounded keyframe anchors for visual guidance, and Metric-Guided Candidate Routing to select the highest-quality videos from heterogeneous generators via multi-dimensional evaluation. Extensive evaluations demonstrate the superiority of our approach, which achieved first place in both Track 1 and Track 2 of the ACM MM 2026 Identity-Preserving Video Generation Challenge, demonstrating its effectiveness across both facial identity preservation and sequential action generation settings.

arXiv:2609.04273v2eess.IV

Scalable Neural Video Representation Compression

Scalable video coding (SVC) encodes a video into a layered bitstream of a base layer and one or more enhancement layers, enabling decoding at different bitrate/quality/resolution operating points for diverse devices and network conditions. SVC has been incorporated into major video coding standards and has recently attracted growing interest for both scene-agnostic and scene-adaptive neural video codecs. Among the latter, implicit neural representation (INR) based codecs overfit a compact neural network to an individual video, offering fast decoding and competitive coding efficiency. However, research on scalable INR-based compression remains in its infancy: existing methods introduce additional network layers, which couple the bitrate with the decoding complexity and cannot match strong scalable/non-scalable codecs. This paper proposes S-NVRC, a scalable INR-based video codec that jointly supports fine-grained bitrate and decoding complexity scalability from a single embedded bitstream. It adopts a coarse-to-fine prefix for feature grids and a nested prefix for network layers, which scale bitrate and decoding complexity, respectively. With a single encoding, S-NVRC spans a wide range of bitrate and decoding complexity and outperforms SHM-12.4 and the multi-layer VTM-20.0 by 43.7% and 5.6% in BD-rate on the UVG dataset.

arXiv:2611.04932v1cs.CV

VISTA: Vision-Language Integrated Semantic Triage Architecture for Online Face Clustering in Long Videos

Clustering faces by identity in long, continuous video streams is a challenging task that serves as a building block for human-centric video analysis. The majority of existing face clustering models are offline, i.e., they assume the entire data to be available altogether. In many practical scenarios, however, the entire data is either unavailable or poses computational challenges to be processed in its entirety. A handful of online methods that exist rely on geometric visual features and simple thresholding techniques, which fail when the faces exhibit significant appearance variations in long, unconstrained videos. To this end, we present VISTA, the first vision language framework for online face clustering in long videos. VISTA leverages semantic reasoning in a controlled, uncertainty aware fashion to maintain both high cluster purity and long range identity continuity. This makes it robust to the variations in illumination, pose, and costume changes that the faces and characters exhibit in long real world videos. Extensive experiments on the VPCD benchmark dataset (with 15 full episodes from The Big Bang Theory [2007-2019], Buffy the Vampire Slayer [1997-2003], and Sherlock [2010-2017]) demonstrate that VISTA achieves a robust precision and recall balance, consistently improving over the strongest matched online unsupervised baseline (COFC) under identical detector, embeddings, and evaluation protocol. Code and supplementary material: https://github.com/SkyJiashu/VISTA.

arXiv:2611.04841v1cs.AI

DGC-Net: Cross-Modal Dynamic Routing with Global-Guided Compensation in Multimodal Sentiment Analysis

User-generated videos, including opinion vlogs and video reviews, provide a rich source of people's attitudes toward products, services, and social topics. Understanding the sentiment expressed by speakers is important for analyzing online opinions, as transcripts alone may not capture the attitude and intensity conveyed through vocal tone and facial expressions. Multimodal sentiment analysis (MSA) addresses this need by integrating textual, acoustic, and visual cues. However, these cues are often subtle, noisy, and inconsistent across modalities. Existing local affective cue modeling methods typically operate within individual modalities and overlook global contexts from other modalities. This may cause cross-modal affective cue inconsistency. Hierarchical global-local fusion methods integrate global semantics with fine-grained local affective cues. However, they commonly determine the relative contributions of the two representations implicitly, preventing the global-local contributions from adapting to the sample-dependent reliability of local cues and resulting in either insufficient use of informative local cues or excessive interference from unreliable ones. To address these issues, we propose DGC-Net, a Dynamic Gating and Global-Guided Compensation Network for MSA. DGC-Net mitigates cross-modal affective cue inconsistency through cross-modal contribution-aware dynamic expert routing. This mechanism uses auxiliary global contexts to guide local expert selection. Moreover, global-guided dynamic compensation fusion alleviates sample-dependent global-local contribution imbalance while preserving global semantic as the primary anchor and adaptively incorporating expert-refined local cues. Experiments on five multimodal benchmark datasets demonstrate the effectiveness and general applicability of DGC-Net, and ablation studies verify the contribution of each proposed component.

arXiv:2611.04591v1cs.CV

SparseGraph-I2P: Consistency-Guided Hypothesis Selection for Efficient Image-to-Point-Cloud Registration

Coarse-to-fine image-to-point-cloud (I2P) registration risks discarding correct but ambiguous patch pairs, while retaining more candidates repeats fine matching over overlapping patches. SparseGraph-I2P aims to improve registration recall by increasing the inlier ratio at the pose solver, using sensor or monocular depth. A Coarse Hypothesis Graph combines appearance with depth-derived first- and second-order compatibility to select a fixed budget of patch pairs. Sparse Fine Matching decodes each queried pixel once and skips pixel--point comparisons covered by an ancestor patch paired with the same point-cloud region. With sensor depth, depth-aware filtering ranks correspondences by SC2 spectral support for a single EPnP solve. With monocular depth, depth or inverse depth is calibrated from correspondence geometry before filtering. SparseGraph-I2P achieves the highest registration recall and lowest latency among the compared methods on RGB-D Scenes V2 and 7-Scenes with sensor depth. With predicted depth, it retains the highest recall on RGB-D Scenes V2 at an additional calibration cost. Code will be released.

arXiv:2611.04111v1cs.CV

Generalizable AI-Generated Image Detection with Embedding-Space Explanations

AI-generated images have reached unprecedented realism, while instruction-driven AI editors re-synthesize images for local changes, leaving no splice boundary to localize. Most detectors are black boxes with unverifiable verdicts and generalize poorly because adapting a backbone on the detection distribution memorizes generator-specific fingerprints that quickly become obsolete. Specialist detectors built on pure-vision representations, particularly DINOv3, generalize well across generators, but explainable methods forgo this advantage: jointly training one multimodal model for classification and language generation subjects its visual representation to the language objective. We therefore require that explanation training cannot alter the verdict. We introduce FOX, which first builds and locks a specialist DINOv3 detector, then uses its representation to guide a lightweight VLM toward forensic explanation. A joint-embedding predictor maps locked visual evidence into a frozen text encoder's semantic space, producing guidance that the VLM verbalizes on demand. FOX achieves 97.4 mean AUC across 14 unseen generators, 69.8 BERTScore on XPlainVerse, and 56.1 on LOKI.

arXiv:2611.03674v1cs.MM

ReVi: Recursive Recap Video Generation

Novels tell rich stories, but their textual form cannot directly present the sights and sounds of the worlds they describe. Film adaptations offer an audiovisual experience but require substantial production effort and viewing time; film recaps are shorter but depend on existing footage. Neither route directly provides a concise audiovisual account of a novel. We therefore study novel-to-recap video generation: producing a self-contained, source-grounded audiovisual retelling directly from a novel within a fixed viewing-time budget. This task requires compressing the story while preserving causal connections, allocating sufficient time to retained events, and aligning narration with newly generated visuals. We propose ReVi, a dual-loop multi-agent framework guided by edit first, generate second. Its Inner Loop plans story content, narration time, and corresponding shots before video generation. Across novels, its Outer Loop performs recursive self-improvement by revising both the editing policy and the meta policy that governs its revision. Experiments on two evaluation sets show that ReVi improves recap quality and duration compliance over representative baselines, with human ratings and multi-round improvement results providing further support.

arXiv:2606.05861v3cs.MM

LLMCodec: Adapting Video Codecs for Efficient Weight Compression of Large Language Models

The rapid development of large language models(LLMs) has led to remarkable advances in natural language processing. However, the increasing scale of these models introduces substantial challenges in terms of storage, transmission, and deployment. Though great efforts have been devoted to model compression and quantization, existing methods often rely on fine-tuning or calibration data, which exhibit limited generalization across different tensor types. In this paper, we argue that video codecs offer a promising solution for LLM compression, due to their inherent compatibility with matrix structured data, configurable compression strategies, and the availability of highly optimized, off-the-shelf implementations. Therefore, we present LLMCodec, a video codec-based LLM compression method that integrates affine quantization with the recent VVC/H.266 video codec. Beyond VVC, we further compare a range of video codecs and encoding profiles to evaluate their impact on compression performance. Experiments on different models demonstrate the robustness and generality of LLMCodec. Notably, on LLaMA-3-8B at 2-bit precision, LLMCodec reduces perplexity by over 1.5x and improves downstream task accuracy by 21% compared with the existing method.

arXiv:2604.01569v3cs.CV

VideoZeroBench: 시공간 증거 검증을 통한 비디오 MLLM의 한계 탐색

VideoZeroBench: Probing the Limits of Video MLLMs with Spatio-Temporal Evidence Verification

비디오 멀티모달 대규모 언어 모델은 기존 벤치마크에서 우수한 성과를 거두고 있으나, 답변의 정확도만으로는 모델이 질문에 답하는 데 필요한 증거를 실제로 찾아낼 수 있는지 입증하지 못합니다. 본 논문에서는 13개 비디오 도메인에 걸쳐 수작업으로 주석을 단 질문-답변 쌍을 포함하는 까다로운 롱비디오(long-video) 벤치마크인 VideoZeroBench를 소개합니다. 질문들은 미세한 단서, 순식간에 지나가는 사건, 여러 구간에 걸쳐 분산된 증거를 목표로 합니다. 해당되는 경우 시간적 구간과 키프레임 바운딩 박스가 주석으로 제공됩니다. 모든 질문은 답변 타당성과 증거 품질에 대해 두 차례의 교차 검증을 거칩니다. 5단계 진단 프로토콜을 통해 증거 힌트 유무에 따른 답변 능력을 비교한 뒤, 답변 정확도와 독립적으로 평가된 시간 및 공간 그라운딩(grounding)을 결합합니다. 평가된 19개 모델 중 표준 QA 정확도가 가장 높은 모델은 Gemini-3.7-Flash로 24.8%(레벨 3)를 기록했습니다. 정답과 정확한 시공간 위치 특정을 동시에 요구할 경우(레벨 5), 1.8%를 초과하는 모델은 없었습니다. 원자적 능력, 증거 범위, 입력 모달리티, 비디오 기반 사고(thinking-with-videos) 추론에 대한 분석은 평가 대상 시스템이 어떤 부분에서 어려움을 겪는지 구체적으로 보여줍니다. 이러한 발견은 롱비디오 질의응답에서 보다 정밀한 증거 탐색 및 위치 특정이 필요함을 시사합니다. 코드와 데이터는 공개되어 있습니다.

arXiv:2611.02599v1cs.HC

Camera-Free Object Discovery and Interaction with GazeTags

We present GazeTags, a camera- and computer-vision-free approach for worn head-gaze object awareness. Our small electronic tags cost roughly \$2 each and can operate for more than a year on a coin cell battery. They work in concert with a user-worn device (smartglasses, hearables, etc.) that periodically emits a focused infrared signal aligned with the user's head direction. Only GazeTags within that narrow head-oriented beam wake and respond, providing their identity and optional payloads such as object state or live sensor readings. By avoiding cameras and computer vision pipelines, our approach is very energy efficient, able to operate continuously even in battery-constrained wearables, and sidesteps non-trivial privacy issues. We describe our hardware design, characterize our system performance and operating tradeoffs, and demonstrate how GazeTags can enable practical gaze-aware environments.

arXiv:2611.01162v1cs.CV

SHield: Watermarking 4D Gaussian Splatting via Spherical Harmonic Coefficient Perturbation

4D Gaussian Splatting (4DGS) models are computationally expensive to train but distributed as easily copied parameter files. Existing watermarks protect these assets by fine-tuning the scene against a neural decoder to extract bit strings from rendered video, a heavy process vulnerable to the temporal dynamics of dynamic scenes. However, if the threat model targets the redistributed file itself, verification can be vastly simplified. Because the canonical Spherical-Harmonic (SH) coefficients of 4DGS are time- and view-invariant, parameter-space perturbations can be recovered via deterministic arithmetic. We propose **SHield**, the first non-blind, decoder-free watermarking approach that natively embeds full-resolution RGB images into 4DGS files. SHield tiles a zero-mean payload across the SH coefficients of a key-selected Gaussian subset, applying a brief anchored optimization to preserve visual fidelity. To verify ownership, it geometrically re-associates the key using canonical Gaussian centres to survive file tampering and recovers the payload in closed form. Across 30 scenes (D-NeRF, HyperNeRF, DyNeRF), SHield guarantees near-perfect payload recovery (Pearson $r \geq 0.991$) on all 150 tested models with a mean rendering penalty of $\leq 0.13$~dB. Held-out evaluation yields perfect detection on 400 true trials and zero false positives across $1{,}600$ wrong-subset null trials. We also strictly bound the method's robustness: SHield survives INT8 quantization and $30\%$ opacity pruning for the image payload, and on aggressive SH-degree-1 truncation bit payloads are still decodable.

arXiv:2611.00192v1eess.IV

VC-1K: A Diverse Video Test Set for Real-Time Communication

We introduce Video Conferencing 1000 (VC-1K), an open-source test set comprising approximately 1,000 video sequences specifically designed for evaluating video processing and encoding pipelines in real-time communication (RTC) scenarios. Unlike existing test sets that primarily rely on studio-produced or screen content, VC-1K consists of user-generated video call content captured on consumer mobile devices under a wide variety of conditions. The dataset predominantly features talking-head sequences with distinct faces, but also includes a small subset of outward-facing, rear-camera videos without faces to reflect additional common RTC use cases. It spans a wide range of real-world variables including device type, lighting conditions, motion levels, indoor/outdoor environments, and demographic diversity. All video sequences are accompanied by synchronous audio to facilitate multimodal research. We detail the dataset's design methodology and statistical distribution, and demonstrate its utility through a comparative encoder evaluation of OpenH264, SVT-AV1, and libaom under constant bitrate (CBR) encoding at resolutions typical of RTC deployments. Furthermore, two super-resolution approaches, RealESRGAN and GFPGAN, are evaluated on VC-1K, highlighting the dataset as a useful benchmark for pre- and post-processing research. VC-1K is intended to serve as a publicly available dataset for the video RTC research community.

arXiv:2611.00069v1cs.CV

SAMP: Self-Adaptive Margin Prioritization For Robust Diffusion Watermarking

Diffusion models have made high-fidelity visual generation widely accessible, while also lowering the barrier to producing convincing visual forgeries and complicating content provenance. Existing watermarking methods typically use a fixed embedding strength throughout denoising or treat all watermark dimensions equally. As a result, they may introduce unnecessary distortion when some watermark dimensions are already reliable, while failing to sufficiently reinforce those that are more vulnerable to subsequent perturbations. Therefore, robust watermarking requires adapting both the embedding strength during denoising and the optimization priority across watermark dimensions. To address this problem, we present SAMP, a self-adaptive framework consisting of global margin regulation and vulnerability-sensitive allocation. Global margin regulation adjusts the embedding strength according to the separation between the target codeword and competing codewords, together with the estimated effects of common attacks. Vulnerability-sensitive allocation identifies watermark dimensions that are more easily weakened by local perturbations and gives them higher optimization priority. Experimental results demonstrate that SAMP achieves an average information recovery accuracy of 0.993. Compared with state-of-the-art methods, SAMP improves PSNR by 16.8% and reduces LPIPS by 47.1%. These results validate the effectiveness of SAMP for reliable and high-quality diffusion watermarking.

arXiv:2611.00031v1cs.CV

Monocular Badminton Speed Estimation with Air Drag and Court Geometry

A single camera cannot distinguish between a slowly moving object nearby and a rapidly moving object in the distance without additional information. For a shuttlecock, air drag provides a strong deceleration cue that can be combined with known court geometry. We fit a gravity-and-drag trajectory to smartphone video and compare a free monocular fit with a fit constrained to cross the net plane. Camera calibration uses court lines, and the trajectory is fitted to observed shuttle positions and frame timestamps by robust nonlinear least squares. We evaluated 51 badminton shots from two recording sessions, spanning drop shots, drives, half-smashes and smashes. The net-plane fit reduced the median absolute speed difference from a drag-based stereo reference from 13.3% to 8.6%. A second reference, obtained by differentiating triangulated positions without a drag model, also gave a smaller median difference for the net-plane fit. The results support combining court geometry with flight dynamics for badminton speed estimation from a single camera.

arXiv:2610.10448v2cs.MM

MemoCare: 자동화된 인지 선별 검사를 위한 대화형 멀티모달 모바일 시스템

MemoCare: An Interactive Multimodal Mobile System for Automated Cognitive Screening

MemoCare is a React Native mobile system for automated multimodal cognitive screening in English and Vietnamese. It integrates speech, temporal and spatial orientation, touchscreen interaction, and visuoconstruction. Google Speech-to-Text transcripts are scored with deterministic, task-specific rules. GPS-grounded current and historical administrative names support location scoring, touch events verify actions, and lightweight convolutional neural networks evaluate drawings. A separate model provides visual interpretation. Software verification passed 151/151 predefined language, spatial-answer, and touch-scoring regression cases and 48/48 fixed-coordinate spatial tests across four countries. Transcript-based tests exclude speech recognition accuracy. On a locked 71-image test set, validation-selected ShuffleNetV2 x1.5 achieved 91.33% mean balanced accuracy and 78.87% exact three-criterion accuracy. Four clinician co-authors rated the workflow (pooled median 4/5 across eight criteria; item medians 3-4.5), highlighting improvements needed in speech handling, response acceptance, and result transparency. The workflow automatically produces item-level and total scores.

arXiv:2606.25547v2cs.CV

공간-주파수 협업 및 비가역 메커니즘을 갖춘 효율적인 교차 스케일 가역 은닉 네트워크

Efficient Cross-Scale Invertible Hiding Network with Spatial-Frequency Collaboration and Non-Invertible Mechanism

기존 가역 신경망(INN) 기반 이미지 은닉 방법들이 단일 스케일·단일 도메인 특징 추출과 제한된 비선형 표현으로 화질 저하를 겪는 문제를 해결하기 위해, CrosInv는 픽셀 셔플과 Haar 웨이블릿 변환을 활용한 교차 스케일 가역 모듈(CIM)과 비가역 교차 밀집 모듈(NCDM)을 결합해 공간-주파수 협업 특징과 향상된 비선형 표현을 동시에 달성한다. COCO, ImageNet, BOSSBase 세 가지 벤치마크 데이터셋에서 수행한 실험을 통해 기존 ISN, HiNet, StegFormer, StarINN 대비 우수한 이미지 품질 및 스테그분석 저항성을 검증하였다.

arXiv:2507.01652v2cs.CV

Autoregressive Image Generation with Linear Complexity: A Spatial-Aware Decay Perspective

Autoregressive (AR) models have recently emerged as a promising paradigm for image generation, as they can model visual tokens through the same next-token prediction principle that underlies large language models. However, most competitive AR image generators still rely on Transformer attention, whose quadratic token interaction cost and growing key-value cache become increasingly expensive for high-resolution visual sequences. Linear attention provides an attractive route to efficient AR generation, but our study shows that directly importing linear or decay-based attention from language modeling leads to clear quality degradation. We identify a key cause: after a 2D token map is flattened by raster scan, 1D sequence locality no longer faithfully reflects spatial locality, and conventional decay mechanisms may introduce a false-locality bias at row transitions. To address this problem, we propose Linear Attention with Spatial-Aware Decay (LASAD), a simple recurrent attention mechanism that modulates decay according to the original 2D token geometry while preserving linear complexity. Built upon this layer, LASADGen replaces standard attention in an AR image generator with spatial-aware linear attention. Experiments on ImageNet demonstrate that LASADGen achieves a favorable quality-efficiency tradeoff, improving generation quality over linear-attention baselines while reducing inference cost and memory usage compared with Transformer-based AR models.

arXiv:2611.02680v1cs.SD

VioTech: Frame-Level Violin Playing Technique Detection with Multi-Scale Mixture-of-Experts

Playing technique is crucial for expressive violin performance, yet few violin technique detection models exist. Current systems mainly rely on note-level pipelines that first estimate note boundaries and then classify techniques. This is especially problematic for bowed string instruments such as the violin, where soft onsets and continuous bow-string interaction render note segmentation unreliable. Here, we propose VioTech, a frame-level violin technique detection framework that predicts techniques directly at 10 ms resolution without explicit note segmentation. To handle the large temporal-scale variation among techniques, VioTech combines a frozen MERT encoder with a multi-scale Mixture-of-Experts (MoE) architecture and task-specific routing for tonal, articulation, and legato-related predictions. We also introduce VioTech120, the first manually annotated dataset of violin playing techniques constructed from real solo performance recordings, as well as an accompanying note-labeling user interface. Results demonstrate that VioTech achieved consistent improvements over prior methods and significantly enhanced cross-dataset generalization, establishing a new state-of-the-art in violin technique-aware transcription.