ECCV 2026 论文列表

This paper proposes the Distribution-Alignment Bridge (DAB), a framework that reconceptualizes text-to-video retrieval as a distribution alignment task rather than traditional deterministic point matching. By modeling both text and video embeddings as Gaussian distributions defined by mean and variance, DAB explicitly accounts for modalityspecific uncertainty. We employ a deterministic, diffusion-inspired bridge to iteratively refine text distributions toward their target video distributions through a truncated refinement process. This approach unifies probabilistic embedding and distributional transformation into a cohesive, end-to-end trainable system. To optimize cross-modal similarity, we introduce a distribution-aware contrastive loss based on Kullback–Leibler divergence. Extensive evaluations on MSR-VTT, MSVD, and VATEX benchmarks confirm that DAB significantly outperforms existing probabilistic and diffusion-based baselines, while providing calibrated uncertainty-aware ranking through bridge-induced distributional margins.

Diffusion transformers have demonstrated remarkable generative capabilities, yet the rich perceptual representations computed across their denoising trajectory are discarded once the content is rendered. We present MMDiff, a framework that transforms a frozen diffusion transformer into a multi-modal generative system that jointly produces images alongside any combination of dense perceptual modalities using lightweight decoder heads. Our central finding is that perceptual information is temporally distributed along the denoising trajectory, and that multi-timestep feature fusion with spatially varying aggregation weights is essential, improving semantic segmentation results by up to 28.7% mIoU over single-timestep extraction. We further adopt conceptdriven attention extraction for interpretable spatial guidance, and show that frozen diffusion features are competitive with and complementary to state-of-the-art encoders such as DINOv3. By training only lightweight decoder heads on a frozen backbone, we achieve strong performance in semantic segmentation, salient object detection, and depth estimation, and demonstrate that this framework enables effective synthetic data generation at scale. Code, models, and results are available here.

Video Super-Resolution (VSR) aims to restore high-quality video frames from low-resolution (LR) estimates, yet most existing VSR approaches behave like black boxes at inference time: users cannot reliably correct unexpected artifacts, but instead can only accept whatever the model produces. In this paper, we propose a novel interactive VSR framework dubbed SparkVSR that makes sparse keyframes a simple and expressive control signal. Specifically, users can first super-resolve or optionally a small set of keyframes using any off-the-shelf image superresolution (ISR) model, then SparkVSR propagates the keyframe priors to the entire video sequence while remaining grounded by the original LR video motion. Concretely, we introduce a keyframe-conditioned latent-pixel two-stage training pipeline that fuses LR video latents with sparsely encoded HR keyframe latents to learn robust cross-space propagation and refine perceptual details. At inference time, SparkVSR supports flexible keyframe selection and a reference-free guidance mechanism that continuously balances keyframe adherence and blind restoration, ensuring robust performance even when reference keyframes are absent or imperfect. Experiments on multiple VSR benchmarks demonstrate improved temporal consistency and strong restoration quality, surpassing baselines by up to 24.6%, 21.8%, and 5.6% on CLIP-IQA, DOVER, and MUSIQ, respectively, enabling controllable, keyframe-driven video super-resolution. Moreover, we demonstrate that SparkVSR is a generic interactive, keyframe-conditioned video processing framework as it can be applied out of the box to unseen tasks such as old-film restoration and video style transfer. Code is available at: https://github.com/tacogroup/SparkVSR.

While multimodal integration significantly improves computer vision models, deploying them incurs prohibitive inference costs and requires scarce, perfectly paired datasets. Recent methods address this data bottleneck by synthesizing missing modalities via generative AI, yet they introduce a severe inefficiency: the Decode-Encode Loop. Specifically, information-rich generative latents are decoded into noisy raw signals, forcing the downstream classifier to waste capacity re-encoding them. To bypass this bottleneck, we propose Direct Latent Augmentation (DLA), utilizing undecoded generative latents directly as privileged information. Furthermore, to transfer this dense knowledge to a purely visual student, we introduce Multilayer Explicit Simulated Synesthesia (MESSy). Instead of enforcing rigid representation matching, which forces the student to distort its native visual features to accommodate complex multimodal topologies, MESSy uses a predictive objective to safely internalize these physical priors. Empirical results demonstrate that our framework significantly outperforms raw data augmentation and traditional distillation. Ultimately, our approach yields highly accurate unimodal students with “synesthetic” latent structures that are inherently aligned with modalities they have never directly observed.

Implicit neural representations (INRs) have recently emerged as a promising approach to video compression, delivering competitive ratedistortion performance alongside rapid decoding. However, existing neural video codecs struggle to balance complexity and scalability. Lightweight models often suffer from degraded compression performance when scaled to different bitrate/quality levels, whereas high-performance models exhibit limited scalability, as their model complexity typically increases with quality. This lack of a unified architecture capable of maintaining consistent complexity across a wide range of bitrates severely limits their diverse real-world deployment. To address these challenges, we introduce NVRC++1, a novel INR-based video codec that utilizes a lightweight INR with multiple high-resolution feature grids, providing high scalability at any given complexity level. This is paired with an optimization framework that enables efficient overfitting on high-resolution grids for long video sequences, thereby exploiting spatio-temporal redundancies without prohibitive computational or memory overhead. Additionally, an advanced entropy model is designed for efficiently compressing the high-dimensional grid parameters. As a result, NVRC++ provides four complexity levels (from 7kMACs/pixel to 360kMACs/pixel), each spanning wide bitrate and quality ranges while supporting real-time decoding. The experimental results show that NVRC++ offers a much faster decoding speed (up to 7.6x) compared to the SOTA INR-based video codec, NVRC, while delivering comparable performance.

Brain tumor MRI segmentation is essential for clinical diagnosis and treatment planning. Although deep learning methods have achieved significant progress, most models operate solely in the image domain and overlook the influence of k-space sampling on spatial structure formation. This physical information gap often limits model generalization when encountering varying scanning conditions across different clinical centers. To address this issue, we propose SPHERE, Sampling-Prior Harmonized rEconstruction-aware Representation lEarning, a samplingaware framework for brain tumor MRI segmentation. SPHERE-Recover estimates k-space sampling–related geometric parameters directly from reconstructed MRI and transforms discrete scanning conditions into a continuous deformation vector field (DVF) that represents samplinginduced structural perturbations. Building on this prior, we develop SPHERE-Seg and introduce a Deformation Prior Module (DPM) to inject the DVF into feature modeling, enabling sampling-driven structural consistency learning. Experiments on BraTS and MSD show that SPHERE improves segmentation accuracy, boundary stability, and crossdataset generalization. These results demonstrate that explicitly modeling MRI sampling mechanisms yields robust, generalizable, and efficient brain tumor segmentation.

Weakly supervised object localization (WSOL) aims to train an object locator using only image-level annotations. Recent progress in WSOL has been predominantly driven by visual transformer architectures, which effectively model long-range feature dependencies through self-attention mechanisms and multilayer perceptrons. However, existing transformer-based approaches typically depend solely on location tokens for localization, neglecting valuable semantic insights provided by class tokens, resulting in diffuse activations and imprecise localization. In this paper, we propose a novel framework named Cross-token Guidance TRansformer (CGTR), which enhances semantic coherence and localization precision by facilitating effective interaction between class and location tokens. Specifically, we introduce an Attention Regulation Module (ARM) to globally refine spatial activations using semantic guidance from class tokens, mitigating contextual biases. Additionally, we develop a Filter Regulation Module (FRM) that applies local structural refinement guided by semantic information, enriching the semantic representation of localization maps. Extensive experimental evaluations demonstrate the effectiveness of the proposed CGTR, achieving substantial and consistent performance gains compared with previous approaches on both the CUB-200-2011 and ILSVRC datasets.

Urban-scale Visual Place Recognition (VPR) aims to identify the geographic location of a query image by matching it against a geo-tagged database. While recent methods achieve impressive performance, they overlook a serious long-tailed problem hidden in urban-scale datasets, which biases the model towards locations with abundant images and ignores less-visited areas, causing models to systematically favor frequently photographed locations while failing in sparsely covered areas. In this paper, we systematically characterize this imbalance challenge and propose Distribution-Aware Place Recognition (DAPR), a model-agnostic plug-in framework that rebalances gradient contributions across head and tail classes. Additionally, within classification-retrieval pipelines, DAPR applies a multi-scale distance search mechanism to compute per-class distributional compactness, providing complementary gains at the retrieval stage. On the large-scale SF-XL benchmark, our framework outperforms the previous classification-retrieval baseline by 18.3% on test set v1, and 6.7% on test set v2. As a plug-in module, it achieves consistent improvements across representative VPR methods on SF-XL, MSLS, and Pitts30k, demonstrating broad generalizability across different methods and benchmarks.

Modern multimodal large language models (MLLMs) achieve impressive results on vision-language benchmarks, yet existing evaluations judge only final answers, making shortcuts indistinguishable from genuine understanding. We introduce CRYSTAL (Clear Reasoning via Yielded Steps, Traceability and Logic), a diagnostic benchmark with 6,372 instances that evaluates multimodal reasoning through verifiable intermediate steps. We propose two complementary metrics: Match F1, which scores step-level precision and recall via semantic similarity matching, and Ordered Match F1, which further penalizes disordered reasoning chains. References are constructed through a Delphi-inspired pipeline where four independent MLLMs generate trajectories, aggregated via semantic clustering and validated through human quality gates. Evaluation of 20 MLLMs, including commercial frontier systems not used during benchmark construction, reveals systematic failures invisible to accuracy: universal cherry-picking (precision far exceeds recall), non-monotonic scaling trade-offs, and disordered reasoning, where even the strongest reasoners preserve under 60% of matched steps in correct order. Beyond evaluation, we propose the Causal Process Reward (CPR), a multiplicative reward that couples answer correctness with step-level alignment, and CPR-Curriculum, which progressively increases reasoning difficulty during training. CPR-Curriculum achieves +32% Match F1 via GRPO where additive reward strategies fail, improving reasoning without manual step annotation.

Video reasoning segmentation demands pixel-accurate object tracking across hundreds of frames under complex natural language queries, producing dense spatiotemporal tokens whose quadratic selfattention cost makes long-video processing prohibitive. Existing methods address this through token compression, yet typically operate on encoder features lacking temporal context, constraining selection before content redundancy can be reliably assessed. Informed compression requires contextual awareness, but acquiring that awareness at full resolution incurs the same quadratic cost compression aims to reduce. State-space models resolve this constraint, as their linear recurrence selectively conditions each token on temporal context at O(T) cost, producing representations where content redundancy becomes assessable. Building on this, Selective SpatioTemporal Aggregation and Compression (STAC) enriches features via decoupled bidirectional spatial and causal temporal scanning, leveraging recurrence-derived redundancy for hierarchical compression with adaptive thresholds optimised with segmentation objective. STAC achieves 85% token reduction and 1.8× speedup while surpassing compression-free baselines on reasoning segmentation benchmarks in a zero-shot streaming-compatible setting. Code is available here.

Wavefront sensing involves estimating the phase and intensity of light, enabling a wide range of imaging applications, from adaptive optics and astronomy to biomedical imaging. Since conventional image sensors can only measure the spatial intensity distribution, phase retrieval arises as the central problem in wavefront sensing. Conventional interferometric approaches like phase-shifting interferometry (PSI) can recover phase information, but they rely on a stable reference beam that is difficult to realize in practical settings. To overcome this limitation, we propose a novel self-reference framework that relies on interference between shifted copies of the incoming wave; this results in pairwise phase differences between shifted pixels. We formulate an analytical solution for the complete phase retrieval based on the propagation of these differences across a connected graph. Furthermore, we provide a theoretical analysis of optimal measurement patterns, proving that co-prime shifts guarantee a connected graph and bound worst-case error accumulation, yielding a provably robust method. Extensive simulations demonstrate that complete phase profiles can be recovered from as few as eight shifted measurements, outperforming several existing approaches. Finally, we validate our framework using a hardware prototype, demonstrating real experiments for optical phase profile recovery, auto-refocusing, and imaging through scattering media.

Creating photorealistic and temporally coherent animatable human avatars from RGB videos remains challenging. Current methods struggle to capture realistic cloth dynamics, producing over-smoothed appearance or severe artifacts on out-of-distribution poses. This limitation stems from a fundamental oversight: existing approaches neglect the temporal causality inherent in cloth physics, where current states emerge from previous states through temporal evolution rather than instantaneous skeletal configurations alone. Without explicit modeling of this causal structure, networks learn pose-appearance correlations instead of motion evolution, leading to poor generalization. We introduce a dual-stream autoregressive framework that explicitly models both observable geometric information and implicit internal state. The geometric stream propagates surface displacement from the previous frame, while the state stream fuses current features with historical states retrieved from a memory bank. Motion-adaptive aggregation handles spatiallyvarying dynamics, and adaptive regularization balances smoothness with flexibility.Experiments on challenging datasets demonstrate significant improvements in rendering quality, temporal consistency, and generalization to motion patterns beyond training distributions, validating that dual-stream temporal modeling enables realistic cloth dynamics.

VisualAutoRegressivemodeling(VAR)haspioneeredacoarse-to-fine multi-scaleautoregressivegenerativeparadigm,demonstratingstrongcapabilities in image generation. However, VAR still suffers from inherent deficiencies in multi-scale representation learning. Specifically, lower scales primarily capture global semantics, while higher scales focus on fine-grained details. Employing a shared architecture across scales induces optimization conflicts. Moreover, due to the causal autoregressive process, inaccurate semantics at early scales can propagate and significantly degrade the final output. To address these issues, we introduce a scale-aware token-routed Mixture of Experts (MoE) architecture, allowing scale-adaptive expert selection, thereby facilitating decoupled representation learning across scales. In addition, we enhance semantic modeling at early scalesbyincorporatingexternalself-supervisedfeatures.Unlikenaivealignment, we analyse and design a residual feature aggregation scheme tailored to the VAR paradigm. Extensive experiments show that our method significantly improves both training efficiency and generation quality. On the ImageNet 256×256 benchmark, our model achieves a superior FID compared to the dense baseline while requiring only half of the default training epochs and a smaller parameter budget, with a merely marginal increase in training cost. Moreover, the performance gap further widens with larger training epochs.

3D Gaussian Splatting (3DGS) and Neural Radiance Fields (NeRF) have demonstrated remarkable capabilities in photo-realistic novel view synthesis. However, their practical adoption is often hindered by substantial storage requirements and limited rendering efficiency, particularly for scenes with repetitive structures. While existing acceleration methods primarily focus on optimizing individual Gaussian primitives or neural network architectures, they fail to address the fundamental redundancy inherent in repetitive content. To overcome this limitation, we introduce InstGS, the first Gaussian instancing-based accelerated rendering framework. To eliminate redundancy at the representation level, we perform gradient-driven cross-frame instance segmentation to group similar Gaussians into reusable components. A shared Gaussian template with instance-specific offsets is optimized to replace all similar instances, yielding substantial memory saving with negligible loss in visual fidelity. Extensive experiments demonstrate that InstGS achieves high-quality, high-frame-rate, and low-memory rendering performance.

Robust medical image segmentation across modalities remains challenging due to severe domain shifts and the lack of targetdomain labels. While diffusion models have been explored for crossdomain generation and augmentation, target-domain conditional diffusion training typically relies on highly noisy pseudo masks; naively conditioning on a single Arg-Max pseudo-label can corrupt diffusion training and downstream segmentation. We propose UPDiff-UDA, a unified UDA framework whose core is an uncertainty-guided training objective for target-domain conditional diffusion. Given an imperfect sourcetrained segmenter, we use its per-pixel softmax distribution to form ranked pseudo-label maps (Arg-Max, Arg-2nd, Arg-3rd, ...). Each map yields a conditional score estimate, and we aggregate them via pixel-wise confidence weighting to obtain an uncertainty-reweighted score for score matching, improving robustness to pseudo-label noise while leveraging alternative plausible labels in uncertain regions. We further provide a theoretical justification showing that confidence-weighted aggregation follows a minimum-MSE convex-combination principle under the segmenterinduced surrogate label distribution. To improve pseudo-condition quality, we also introduce a feature-guided, low-degree-of-freedom Bézier curve adaptation to reduce appearance gaps. Experiments on multiple public datasets and modality shifts show that UPDiff-UDA generates highfidelity labeled target-style samples for augmentation and consistently outperforms strong UDA baselines. The code for this project is available at: https://github.com/superlc1995/Multi-Channel-UncertaintyDiffusion-UDA

Multimodal Large Language Models (MLLMs) often struggle with fine-grained perception, such as identifying small objects in high-resolution images or detecting key moments in long videos. Existing methods typically rely on complex, task-specific fine-tuning, which reduces generalizability and increases system complexity. In this work, we propose an effective, training-free framework that uses an MLLM’s intrinsic uncertainty as proactive guidance. Our core insight is that a model’s uncertainty decreases when provided with relevant visual information. We introduce a unified mechanism that scores candidate visual inputs by response uncertainty, enabling the model to autonomously focus on the most informative data. We apply this simple principle to three challenging visual tasks: Visual Search, Long Video Understanding, and Temporal Grounding, allowing off-the-shelf MLLMs to achieve performance competitive with specialized, fine-tuned systems. Our results demonstrate that leveraging intrinsic uncertainty is a powerful strategy for improving fine-grained multimodal performance. Code is available at https://github.com/ExplainableML/ug-framework.

In this work, we study Cooperative Spatial Intelligence, the ability of decentralized embodied agents to coordinate effectively under dynamic environmental constraints across city-scale outdoor domains. We introduce Sentinel Challenge, a benchmark where multiple decentralized embodied agents must communicate in natural language to agree on a mutually safe and convenient meeting point within large, city-scale outdoor environments. Each agent must then navigate safely while avoiding dynamic sentinels patrolling the area, with a tool providing coarse spatial information. To address this, we propose CoSaR (Cooperative Spatial Reasoning and Planning), a framework that bridges the high-level communication and planning abilities of foundation models with the precision of classical spatial navigation algorithms. CoSaR enables agents to exchange situational updates, reason over evolving spatial constraints, and collaboratively replan trajectories. Evaluated across 14 city-level scenes with 3–5 agents, CoSaR consistently leads to faster gathering, shorter path lengths, and improved safety. Our results demonstrate that integrating dynamic communication with spatial reasoning is essential for robust multi-agent cooperation. By formalizing this new setting and providing a scalable benchmark, we aim to build a foundation for advancing cooperative spatial intelligence in embodied multi-agent systems. Code and challenge are available at https://github.com/UMass-Embodied-AGI/Sentinel.

In this paper, we study depth perception of vision-language models (VLMs) to isolate the effects of pictorial depth cues and disentangle vision and language influences on model performance. To this end, we combine depth-ordering and odd-one-out psychophysical tasks: the VLMs are presented with images where one object is at different depth relative to other, otherwise identical, objects, and must determine whether the odd-one-out target is closer or farther to the observer. To create stimuli, we generate 2D views from simulated and real 3D scenes while controlling the presence of individual pictorial depth cues, enabling a fine-grained analysis of cue-level contributions. Language effects are examined by varying referring expression clarity. We also introduce a novel metric to quantify vision-vs-language sensitivities. Applying this methodology, we create the Odd-One-Out Depth (O3-D) dataset with 37K real and synthetic images and 147K image-question pairs. Evaluation of 12 open-source and commercial models on O3-D shows under-utilization of depth cues and depth-ordering accuracies between 47% and 56%, with no model above chance level. At the same time, our metric reveals strong linguistic bias in the answers. Neither chain-of-thought (CoT) nor incontext learning (ICL) significantly improves performance, suggesting that static image data alone may be insufficient for depth understanding. All code, the image generation pipeline, and the O3-D dataset are publicly released at https://github.com/lyiqian/o3-d.

We propose a step-by-step video-to-audio (V2A) generation method that provides finer control over the generation process and more realistic audio synthesis. Inspired by traditional Foley workflows, our approach enables incremental generation of complementary sounds, allowing users to author multiple sound events induced by a video. To avoid the need for costly multi-reference video-audio datasets, each generation step is formulated as a negatively guided V2A process that discourages duplication of sounds already present in previously generated tracks. The guidance model is trained by finetuning a pre-trained V2A model on audio pairs from non-overlapping segments of the same video, encouraging it to leverage acoustic context while remaining visually grounded, and enabling training with standard single-reference audiovisual datasets. Objective and subjective evaluations demonstrate that our method enhances the separability of generated sounds at each step and improves the overall quality of the final composite audio, outperforming existing baselines. Our project page is available at: https://ahykw.github.io/sbsv2a/.