CVPR 2025 Papers

In the Blind Image Quality Assessment (BIQA) field, accurately assessing the quality of authentically distorted images presents a substantial challenge due to the diverse distortion types in natural settings. Existing state-of-the-art IQA methods mix a sequence of distortions into entire images to establish global distortion priors, but are inadequate for authentic images with spatially varied distortions. To address this, we introduce a novel IQA framework that employs knowledge distillation tailored to perceive spatially heterogeneous distortions, enhancing quality-distortion awareness. Specifically, we introduce a novel Block-wise Degradation Modelling approach that applies distinct distortions to different spatial blocks of an image, thereby expanding local distortion priors. Following this, we present a Block-wise Aggregation and Filtering module that enables fine-grained attention to the quality information within different distortion areas of the image. Furthermore, to effectively capture the complex relationships between distortions across different regions while preserving overall quality perception, we introduce Contrastive Knowledge Distillation to enhance the model's ability to discriminate between different types of distortions and Affinity Knowledge Distillation to model the correlation among distortions in different regions. Extensive experiments on standard BIQA datasets demonstrate the effectiveness and competitiveness of the proposed method.

Depth completion involves predicting dense depth maps from sparse LiDAR inputs, a critical task for applications such as autonomous driving and robotics. However, sparse depth annotations from sensors limit the availability of dense supervision, which is necessary for learning detailed geometric features. To overcome this limitation, we propose a two-stage knowledge distillation framework that leverages powerful monocular foundation models to provide dense supervision for depth completion. In the first stage, we introduce a pre-training strategy that generates diverse training data from natural images to distill geometric knowledge to depth completion. Specifically, we simulate LiDAR scans by utilizing monocular depth and mesh reconstruction, thereby creating training data without requiring ground-truth depth. Nonetheless, monocular depth estimation suffers from inherent scale ambiguity in real-world settings. To address this, in the second stage, we employ a scale- and shift-invariant loss (SSI Loss) to learn real-world scales when fine-tuning on real-world datasets. Our two-stage distillation framework enables depth completion models to harness the strengths of monocular foundation models. Experimental results show that models trained with our two-stage distillation framework achieve top-ranked performance on the KITTI benchmark, demonstrating improvements in both quantitative and qualitative metrics.

The integration of multimodal data including pathology images and gene profiles is widely applied in precise survival prediction. Despite recent advances in multimodal survival models, collecting complete modalities for multimodal fusion still poses a significant challenge, hindering their application in clinical settings. Current approaches tackling incomplete modalities often fall short, as they typically compensate for only a limited part of the knowledge of missing modalities. To address this issue, we propose a Distilled Prompt Learning framework (DisPro) to utilize the strong robustness of Large Language Models (LLMs) to missing modalities, which employs two-stage prompting for compensation of comprehensive information for missing modalities. In the first stage, Unimodal Prompting (UniPro) distills the knowledge distribution of each modality, preparing for supplementing modality-specific knowledge of the missing modality in the subsequent stage. In the second stage, Multimodal Prompting (MultiPro) leverages available modalities as prompts for LLMs to infer the missing modality, which provides modality-common information. Simultaneously, the unimodal knowledge acquired in the first stage is injected into multimodal inference to compensate for the modality-specific knowledge of the missing modality. Extensive experiments covering various missing scenarios demonstrated the superiority of the proposed method. The code is available at https://github.com/Innse/DisPro.

Active Real-time interaction with video LLMs introduces a new paradigm for human-computer interaction, where the model not only understands user intent but also responds while continuously processing streaming video on the fly. Unlike offline video LLMs, which analyze the entire video before answering questions, active real-time interaction requires three capabilities: 1) Perception: real-time video monitoring and interaction capturing. 2) Decision: raising proactive interaction in proper situations, 3) Reaction: continuous interaction with users. However, inherent conflicts exist among the desired capabilities. The Decision and Reaction require a contrary Perception scale and grain, and the autoregressive decoding blocks the real-time Perception and Decision during the Reaction. To unify the conflicted capabilities within a harmonious system, we present Dispider, a solution built on a Disentangled Perception, Decision, and Reaction framework. Dispider features a lightweight Proactive Streaming Video Processing module that tracks the video stream and identifies optimal moments for interaction. Once the interaction is triggered, an asynchronous Precise Interaction module provides detailed responses, while the processing module continues to monitor the video in the meantime. Our disentangled and asynchronous design ensures timely, contextually accurate, and computationally efficient responses, making Dispider ideal for active real-time interaction for long-duration video streams. Experiments prove that Dispider outperforms existing methods not only in its superior understanding of video content in conventional video QA settings, but also in proactive response capability and temporal awareness under the streaming setting.

DiskVPS is a novel and robust vanishing point (VP) detection scheme based on Hough Transform (HT) over an image-plane-mapped disk region. In the DiskVPS, the image plane is first mapped into a disk region, which is then partitioned into concentrical ring-shaped subspaces. Each subspace is further partitioned into approximately equal-probability Hough cells. By using individual edges rather than edge pairs as voters, DiskVPS can achieve high accuracy with extreme efficiency. Involving no calibration parameters, DiskVPS is particularly suitable for VP detection in uncalibrated images and thus can be applied in image calibration. A comparative experimental study demonstrates that the basic DiskVPS model without parameter optimization achieved significantly better performance over the SOTA in detection accuracy and processing speed with real-world images. The study also shows that DiskVPS is robust against parameter changes.

State-of-the-art machine learning systems are vulnerable to small perturbations to their input, where _small_ is defined according to a threat model that assigns a positive threat to each perturbation. Most prior works define a task-agnostic, isotropic, and global threat, like the l_p norm, where the magnitude of the perturbation fully determines the degree of the threat and neither the direction of the attack nor its position in space matter. However, common corruptions in computer vision, such as blur, compression, or occlusions, are not well captured by such treat models. This paper proposes a novel threat model called \texttt Projected Displacement (PD) to study robustness beyond existing isotropic and global threat models. The proposed threat model measures the threat of a perturbation via its alignment with _unsafe directions_, defined as directions in the input space along which a perturbation of sufficient magnitude changes the ground truth class label. Unsafe directions are identified locally for each input based on observed training data. In this way, the PD-threat model exhibits anisotropy and locality. The PD-threat model is computationally efficient and can be easily integrated into existing robustness pipelines. Experiments on Imagenet-1k data indicate that, for any input, the set of perturbations with small PD threat includes _safe_ perturbations of large l_p norm that preserve the true label, such as noise, blur and compression, while simultaneously excluding _unsafe_ perturbations that alter the true label. Unlike perceptual threat models based on embeddings of large-vision models, the PD-threat model can be readily computed for arbitrary classification tasks without pre-training or finetuning. Further additional task information such as sensitivity to image regions or concept hierarchies can be easily integrated into the assessment of threat and thus the PD threat model presents practitioners a flexible, task-driven threat specification that alleviates the limitations of l_p-threat models.

The parameter-efficient adaptation of the image-text pretraining model CLIP for video-text retrieval is a prominent area of research. While CLIP is focused on image-level vision-language matching, video-text retrieval demands comprehensive understanding at the video level. Three key discrepancies emerge in the transfer from image-level to video-level: vision, language, and alignment. However, existing methods mainly focus on vision while neglecting language and alignment. In this paper, we propose Discrepancy Reduction in Vision, Language, and Alignment (DiscoVLA), which simultaneously mitigates all three discrepancies. Specifically, we introduce Image-Video Features Fusion to integrate image-level and video-level features, effectively tackling both vision and language discrepancies. Additionally, we generate pseudo image captions to learn fine-grained image-level alignment. To mitigate alignment discrepancies, we propose Image-to-Video Alignment Distillation, which leverages image-level alignment knowledge to enhance video-level alignment. Extensive experiments demonstrate the superiority of our DiscoVLA. In particular, on MSRVTT with CLIP (ViT-B/16), DiscoVLA outperforms previous methods by 2.2% R@1 and 7.5% R@sum. The code is available at https://github.com/LunarShen/DsicoVLA.

Story visualization, the task of creating visual narratives from textual descriptions, has seen progress with text-to-image generation models. However, these models often lack effective control over character appearances and interactions, particularly in multi-character scenes. To address these limitations, we propose a new task: customized manga generation and introduce DiffSensei, an innovative framework specifically designed for generating manga with dynamic multi-character control. DiffSensei integrates a diffusion-based image generator with a multimodal large language model (MLLM) that acts as a text-compatible identity adapter. Our approach employs masked cross-attention to seamlessly incorporate character features, enabling precise layout control without direct pixel transfer. Additionally, the MLLM-based adapter adjusts character features to align with panel-specific text cues, allowing flexible adjustments in character expressions, poses, and actions. We also introduce MangaZero, a large-scale dataset tailored to this task, containing 43,264 manga pages and 427,147 annotated panels, supporting the visualization of varied character interactions and movements across sequential frames. Extensive experiments demonstrate that DiffSensei outperforms existing models, marking a significant advancement in manga generation by enabling text-adaptable character customization. The code, model, and dataset will be open-sourced to the community.

Hallucinations in Large Vision-Language Models (LVLMs) significantly undermine their reliability, motivating researchers to explore the causes of hallucination. However, most studies primarily focus on the language aspect rather than the visual. In this paper, we address how LVLMs process visual information and whether this process causes hallucination. Firstly, we use the attention lens to identify the stages at which LVLMs handle visual data, discovering that the middle layers are crucial. Moreover, we find that these layers can be further divided into two stages: "visual information enrichment" and "semantic refinement" which respectively propagate visual data to object tokens and interpret it through text. By analyzing attention patterns during the visual information enrichment stage, we find that real tokens consistently receive higher attention weights than hallucinated ones, serving as a strong indicator of hallucination. Further examination of multi-head attention maps reveals that hallucination tokens often result from heads interacting with inconsistent objects. Based on these insights, we propose a simple inference-time method that adjusts visual attention by integrating information across various heads. Extensive experiments demonstrate that this approach effectively mitigates hallucinations in mainstream LVLMs without additional training costs.

This paper presents a novel approach for joint point-feature detection and tracking, designed specifically for Pixel Processor Array (PPA) vision sensors. Instead of standard pixels, PPA sensors consist of thousands of "pixel-processors", enabling massive parallel computation of visual data at the point of light capture. Our approach performs all computation entirely in-pixel, meaning no raw image data need ever leave the sensor for external processing. We introduce a Descriptor-In-Pixel paradigm, in which a feature descriptor is held within the memory of each pixel-processor. The PPA's architecture enables the response of every processor's descriptor, upon the current image, to be computed in parallel. This produces a "descriptor response map", which, by generating the correct layout of descriptors across the pixel-processors, can be used for both point-feature detection and tracking. This reduces sensor output to just sparse feature locations and descriptors, read-out via an address-event interface, giving a greater than 1000X reduction in data transfer compared to raw image output. The sparse readout and complete utilization of all pixel-processors makes our approach very efficient. Our implementation upon the SCAMP-7 PPA prototype runs at over 3000 FPS (Frames Per Second), tracking point-features reliably under violent motion. This is the first work performing point-feature detection and tracking entirely in-pixel.

Large-scale pre-trained vision models are becoming increasingly prevalent, offering expressive and generalizable visual representations that benefit various downstream tasks. Recent studies on the emergent properties of these models have revealed their high-level geometric understanding, in particular in the context of depth perception. However, it remains unclear how depth perception arises in these models without explicit depth supervision provided during pre-training. To investigate this, we examine whether the monocular depth cues, similar to those used by the human visual system, emerge in these models. We introduce a new benchmark, DepthCues, designed to evaluate depth cue understanding, and present findings across 20 diverse and representative pre-trained vision models. Our analysis shows that human-like depth cues emerge in more recent larger models. We also explore enhancing depth perception in large vision models by fine-tuning on DepthCues, and find that even without dense depth supervision, this improves depth estimation. To support further research, our benchmark and evaluation code will be made publicly available for studying depth perception in vision models.

In recent years, large transformer-based video encoder models have greatly advanced state-of-the-art performance on video classification tasks. However, these large models typically process videos by averaging embedding outputs from multiple clips over time to produce fixed-length representations. This approach fails to account for a variety of time-related features, such as variable video durations, chronological order of events, and temporal variance in feature significance. While methods for temporal modeling do exist, they often require significant architectural changes and expensive retraining, making them impractical for off-the-shelf, fine-tuned large encoders. To overcome these limitations, we propose DejaVid, an encoder-agnostic method that enhances model performance without the need for retraining or altering the architecture. Our framework converts a video into a variable-length temporal sequence of embeddings (TSE). A TSE naturally preserves temporal order and accommodates variable video durations. We then learn per-timestep, per-feature weights over the encoded TSE frames, allowing us to account for variations in feature importance over time. We introduce a new neural network architecture inspired by traditional time series alignment algorithms for this learning task. Our evaluation demonstrates that DejaVid substantially improves the performance of a state-of-the-art large encoder, achieving leading Top-1 accuracy of 77.2% on Something-Something V2, 89.1% on Kinetics-400, and 88.6% on HMDB51, while adding fewer than 1.8% additional learnable parameters and requiring less than 3 hours of training time. Our code is available at https://github.com/darrylho/DejaVid.

All-in-one image restoration aims to recover clear images from various degradation types and levels with a unified model. Nonetheless, the significant variations among degradation types present challenges for training a universal model, often resulting in task interference, where the gradient update directions of different tasks may diverge due to shared parameters. To address this issue, motivated by the routing strategy, we propose DFPIR, a novel all-in-one image restorer that introduces Degradation-aware Feature Perturbations(DFP) to adjust the feature space to align with the unified parameter space. In this paper, the feature perturbations primarily include channel-wise perturbations and attention-wise perturbations. Specifically, channel-wise perturbations are implemented by shuffling the channels in high-dimensional space guided by degradation types, while attention-wise perturbations are achieved through selective masking in the attention space. To achieve these goals, we propose a Degradation-Guided Perturbation Block (DGPB) to implement these two functions, positioned between the encoding and decoding stages of the encoder-decoder architecture. Extensive experimental results demonstrate that DFPIR achieves state-of-the-art performance on several all-in-one image restoration tasks including image denoising, image dehazing, image deraining, motion deblurring, and low-light image enhancement. Our codes are available at https://github.com/TxpHome/DFPIR.

In the field of 3D medical imaging, accurately extracting and representing the blood vessels with curvilinear structures holds paramount importance for clinical diagnosis. Previous methods have commonly relied on discrete representation like mask, often resulting in local fractures or scattered fragments due to the inherent limitations of the per-pixel classification paradigm. In this work, we introduce DeformCL, a new continuous representation based on Deformable Centerlines, where centerline points act as nodes connected by edges that capture spatial relationships. Compared with previous representations, DeformCL offers three key advantages: natural connectivity, noise robustness, and interaction facility. We present a comprehensive training pipeline structured in a cascaded manner to fully exploit these favorable properties of DeformCL. Extensive experiments on four 3D vessel segmentation datasets demonstrate the effectiveness and superiority of our method. Furthermore, the visualization of curved planar reformation images validates the clinical significance of the proposed framework.

Recovering high-quality depth maps from compressed sources has gained significant attention due to the limitations of consumer-grade depth cameras and the bandwidth restrictions during data transmission. However, current methods still suffer from two challenges. First, bit-depth compression produces a uniform depth representation in regions with subtle variations, hindering the recovery of detailed information. Second, densely distributed random noise reduces the accuracy of estimating the global geometric structure of the scene. To address these challenges, we propose a novel framework, termed geometry-decoupled network (GDNet), for compressed depth map super-resolution that decouples the high-quality depth map reconstruction process by handling global and detailed geometric features separately. To be specific, we propose the fine geometry detail encoder (FGDE), which is designed to aggregate fine geometry details in high-resolution low-level image features while simultaneously enriching them with complementary information from low-resolution context-level image features. In addition, we develop the global geometry encoder (GGE) that aims at suppressing noise and extracting global geometric information effectively via constructing compact feature representation in a low-rank space. We conduct experiments on multiple benchmark datasets, demonstrating that our GDNet significantly outperforms current methods in terms of geometric consistency and detail recovery. In the ECCV 2024 AIM Compressed Depth Upsampling Challenge, our solution won the 1st place award. Our codes are available at: https://github.com/Ian0926/GDNet.

Multimodal Large Language Models (MLLMs) excel in various tasks, yet often struggle with modality bias, tending to rely heavily on a single modality or prior knowledge when generating responses. In this paper, we propose a debiased preference optimization dataset, RLAIF-V-Bias, and introduce a Noise-Aware Preference Optimization (NAPO) algorithm. Specifically, we first construct the dataset by introducing perturbations to reduce the informational content of certain modalities, prompting the model to overly rely on a specific modality when generating responses. To address the inevitable noise in automatically constructed data, we combine the noise-robust Mean Absolute Error (MAE) with the Binary Cross-Entropy (BCE) in Direct Preference Optimization (DPO) using a negative Box-Cox transformation and dynamically adjust the algorithm's noise robustness based on the evaluated noise levels in the data.Extensive experiments validate our approach, demonstrating not only its effectiveness in mitigating modality bias but also its significant role in minimizing hallucinations.

3D Gaze estimation is a challenging task due to two main issues. First, existing methods focus on analyzing dense features (e.g., large pixel regions), which are sensitive to local noise (e.g., light spots, blurs) and result in increased computational complexity. Second, an eyeball model can correspond multiple gaze directions, and the entangled representation between gazes and models increases the learning difficulty. To address these issues, we propose De\textsuperscript 2Gaze , a lightweight and accurate model-aware 3D gaze estimation method. In De\textsuperscript 2 Gaze, we introduce two key innovations for deformable and decoupled representation learning. Specifically, first, we propose a deformable sparse attention mechanism that can adapt sparse sampling points to attention areas to avoid local noise influences. Second, we propose a spatial decoupling network with a dual-branch decoding architecture to disentangle invariant (e.g., eyeball radius, position) and variable (e.g., gaze, pupil, iris) features from the latent space. Compared to existing methods, De\textsuperscript 2 Gaze requires fewer sparse features, and achieves faster convergence speed, lower computational complexity, and higher accuracy in 3D gaze estimation.Qualitative and quantitative experiments demonstrate that De\textsuperscript 2 Gaze achieves state-of-the-art accuracy and high-quality semantic segmentation for 3D gaze estimation on the TEyeD dataset.

This paper presents Deformable Neural Vessel Representations (DeNVeR), an unsupervised approach for vessel segmentation in X-ray angiography videos without annotated ground truth. DeNVeR utilizes optical flow and layer separation techniques, enhancing segmentation accuracy and adaptability through test-time training. Key contributions include a novel layer separation bootstrapping technique, a parallel vessel motion loss, and the integration of Eulerian motion fields for modeling complex vessel dynamics. A significant component of this research is the introduction of the XACV dataset, the first X-ray angiography coronary video dataset with high-quality, manually labeled segmentation ground truth. Extensive evaluations on both XACV and CADICA datasets demonstrate that DeNVeR outperforms current state-of-the-art methods in vessel segmentation accuracy and generalization capability while maintaining temporal coherency. Please see our project page at kirito878.github.io/DeNVeR.

Self-supervised learning (SSL) is pervasively exploited in training high-quality upstream encoders with a large amount of unlabeled data. However, it is found to be susceptible to backdoor attacks merely via polluting a small portion of training data. The victim encoders associate triggered inputs with target embeddings, e.g., mapping a triggered cat image to an airplane embedding, such that the downstream tasks inherit unintended behaviors when the trigger is activated. Emerging backdoor attacks have shown great threats across different SSL paradigms such as contrastive learning and CLIP, yet limited research is devoted to defending against such attacks, and existing defenses fall short in detecting advanced stealthy backdoors. To address the limitations, we propose a novel detection mechanism, DeDe, which detects the activation of backdoor mappings caused by triggered inputs on victim encoders. Specifically, DeDe trains a decoder for any given SSL encoder using an auxiliary dataset (which can be out-of-distribution or even slightly poisoned), so that for any triggered input that misleads the encoder into the target embedding, the decoder generates an output image significantly different from the input. DeDe leverages the discrepancy between the input and the decoded output to identify potential backdoor misbehavior during inference. We empirically evaluate DeDe on both contrastive learning and CLIP models against various types of backdoor attacks. Our results demonstrate promising detection effectiveness over various advanced attacks and superior performance compared over state-of-the-art detection methods.