CVPR 2025 论文列表

Instruction-based image editing aims to modify specific image elements with natural language instructions. However, current models in this domain often struggle to execute complex user instructions accurately, as they are trained on low-quality data with limited editing types. We present AnyEdit, a comprehensive multi-modal instruction editing dataset, comprising 2.5 million high-quality editing pairs spanning over 20 editing types and five domains. We ensure the diversity and quality of the AnyEdit collection through three aspects: initial data diversity, adaptive editing process, and automated selection of editing results. Using the dataset, we further train a novel AnyEdit Stable Diffusion with task-aware routing and learnable task embedding for unified image editing. Comprehensive experiments on three benchmark datasets show that AnyEdit consistently boosts the performance of diffusion-based editing models. This presents prospects for developing instruction-driven image editing models that support human creativity.

Large Vision-Language Models (LVLMs) have achieved impressive results across various cross-modal tasks. However, hallucinations, i.e., the models generating counterfactual responses, remain a challenge. Though recent studies have attempted to alleviate object perception hallucinations, they focus on the models' response generation, and overlook the task question itself. This paper discusses the vulnerability of LVLMs in solving counterfactual presupposition questions (CPQs), where the models are prone to accept the presuppositions of counterfactual objects and produce severe hallucinatory responses. To this end, we introduce "Antidote", a unified, synthetic data-driven post-training framework for mitigating both types of hallucination above. It leverages synthetic data to incorporate factual priors into questions to achieve self-correction, and decouples the mitigation process into a preference optimization problem. Furthermore, we construct "CP-Bench", a novel benchmark to evaluate LVLMs' ability to correctly handle CPQs and produce factual responses. Applied to the LLaVA series, Antidote can simultaneously enhance performance on CP-Bench by over 50%, POPE by 1.8-3.3%, and CHAIR & SHR by 30-50%, all without relying on external supervision from stronger LVLMs or human feedback and without introducing noticeable catastrophic forgetting issues.

We propose a unified approach for video-controlled generation, enabling text-based guidance and manual annotations to control the generation of videos, similar to camera direction guidance. Specifically, we designed a two-stage algorithm. In the first stage, we convert all control information into frame-by-frame motion flows. In the second stage, we use these motion flows as guidance to control the final video generation. Additionally, to reduce instability in the generated videos caused by large motion variations (such as those from camera movement, object motion, or manual inputs), which can result in flickering or the intermittent disappearance of objects, we transform the temporal feature computation in the video model into frequency-domain feature computation. This is because frequency-domain signals better capture the essential characteristics of an image, and by ensuring consistency in the video's frequency-domain features, we can enhance temporal coherence and reduce flickering in the final generated video.

Video moment retrieval aims to locate specific moments from a video according to the query text. This task presents two main challenges: i) aligning the query and video frames at the feature level, and ii) projecting the query-aligned frame features to the start and end boundaries of the matching interval. Previous work commonly involves all frames in feature alignment, easy to cause aligning irrelevant frames with the query. Furthermore, they forcibly map visual features to interval boundaries but ignoring the information gap between them, yielding suboptimal performance. In this study, to reduce distraction from irrelevant frames, we designate an anchor frame as that with the maximum query-frame relevance measured by the established Vision-Language Model. Via similarity comparison between the anchor frame and the others, we produce a semantically compact segment around the anchor frame, which serves as a guide to align features of query and related frames. We observe that such a feature alignment will make similarity cohesive between target frames, which enables us to predict the interval boundaries by a single point detection in the 2D semantic similarity space of frames, thus well bridging the information gap between frame semantics and temporal boundaries. Experimental results across various datasets demonstrate that our approach significantly improves the alignment between queries and video frames while effectively predicting temporal moment boundaries. Especially, on QVHighlights Test and ActivityNet Captions datasets, our proposed approach achieves 3.8% and 7.4% respectively higher than current state-of-the-art R1@.7 performance. The code is available at https://github.com/ExMorgan-Alter/AFAFSGD.

Precipitation nowcasting involves using current radar observation sequences to predict future radar sequences and determine future precipitation distribution, which is crucial for disaster warning, traffic planning, and agricultural production. Despite numerous advancements, challenges persist in accurately predicting both the location and intensity of precipitation, as these factors are often interdependent, with complex atmospheric dynamics and moisture distribution causing position and intensity changes to be intricately coupled. Inspired by the fact that in the frequency domain, phase variations are shown to correspond to changes in the position of precipitation, while amplitude variations are linked to intensity changes, we propose an amplitude-phase disentanglement model called AlphaPre, which separately learn the position and intensity changes of precipitation. AlphaPre comprises three key components: a phase network, an amplitude network, and an AlphaMixer. The phase network captures positional changes by learning phase variations, while the amplitude network models intensity changes by alternating between the frequency and spatial domains. The AlphaMixer then integrates these components to produce a refined precipitation forecast. Extensive experiments on four datasets demonstrate the effectiveness and superiority of our method over state-of-the-art approaches. Our code is publicly available at https://github.com/linkenghong/AlphaPre.

Existing Large Multimodal Models (LMMs) generally focus on only a few regions and languages. As LMMs continue to improve, it is increasingly important to ensure they understand cultural contexts, respect local sensitivities, and support low-resource languages, all while effectively integrating corresponding visual cues. In pursuit of culturally diverse global multimodal models, our proposed All Languages Matter Benchmark (ALM-bench) represents the largest and most comprehensive effort to date for evaluating LMMs across 100 languages. ALM-bench challenges existing models by testing their ability to understand and reason about culturally diverse images paired with text in various languages, including many low-resource languages traditionally underrepresented in multimodal research. The benchmark offers a robust and nuanced evaluation framework featuring various question formats, including True/False, multiple choice, and open-ended questions, which are further divided into short and long-answer categories. ALM-bench design ensures a comprehensive assessment of a model's ability to handle varied levels of difficulty in visual and linguistic reasoning. To capture the rich tapestry of global cultures, ALM-bench carefully curates content from 13 distinct cultural aspects, ranging from traditions and rituals to famous personalities and celebrations. Through this, ALM-bench not only provides a rigorous testing ground for state-of-the-art open and closed-source LMMs but also highlights the importance of cultural and linguistic inclusivity, encouraging the development of models that can serve diverse global populations effectively. Our benchmark will be publicly released.

Medical Visual Question Answering (Med-VQA) is a challenging task that requires a deep understanding of both medical images and textual questions. Although recent works leveraging Medical Vision-Language Pre-training (Med-VLP) have shown strong performance on the Med-VQA task, there is still no unified solution for modality alignment, and the issue of hard negatives remains under-explored. Additionally, commonly used knowledge fusion techniques for Med-VQA may introduce irrelevant information. In this work, we propose a framework to address these challenges through three key contributions: (1) a unified solution for heterogeneous modality alignments across multiple levels, modalities, views, and stages, leveraging methods such as contrastive learning and optimal transport theory; (2) a hard negative mining method that employs soft labels for multi-modality alignments and enforces the hard negative pair discrimination; and (3) a Gated Cross-Attention Module for Med-VQA that integrates the answer vocabulary as prior knowledge and select relevant information from it. Our framework outperforms the previous state-of-the-art on widely used Med-VQA datasets like RAD-VQA, SLAKE, PathVQA and VQA-2019. The code will be publicly available.

Cross-modal alignment is crucial for multimodal representation fusion due to the inherent heterogeneity between modalities. While Transformer-based methods have shown promising results in modeling inter-modal relationships, their quadratic computational complexity limits their applicability to long-sequence or large-scale data. Although recent Mamba-based approaches achieve linear complexity, their sequential scanning mechanism poses fundamental challenges in comprehensively modeling cross-modal relationships. To address this limitation, we propose AlignMamba, an efficient and effective method for multimodal fusion. Specifically, grounded in Optimal Transport, we introduce a local cross-modal alignment module that explicitly learns token-level correspondences between different modalities. Moreover, we propose a global cross-modal alignment loss based on Maximum Mean Discrepancy to implicitly enforce the consistency between different modal distributions. Finally, the unimodal representations after local and global alignment are passed to the Mamba backbone for further cross-modal interaction and multimodal fusion. Extensive experiments on complete and incomplete multimodal fusion tasks demonstrate the effectiveness and efficiency of the proposed method. For instance, on the CMU-MOSI dataset, AlignMamba improves classification accuracy by 0.9%, reduces GPU memory usage by 20.3%, and decreases inference time by 83.3%.

Vision-Language Models (VLMs) bring powerful understanding and reasoning capabilities to multimodal tasks. Meanwhile, the great need for capable aritificial intelligence on mobile devices also arises, such as the AI assistant software. Some efforts try to migrate VLMs to edge devices to expand their application scope. Simplifying the model structure is a common method, but as the model shrinks, the trade-off between performance and size becomes more and more difficult. Knowledge distillation (KD) can help models improve comprehensive capabilities without increasing size or data volume. However, most of the existing large model distillation techniques only consider applications on single-modal LLMs, or only use teachers to create new data environments for students. None of these methods take into account the distillation of the most important cross-modal alignment knowledge in VLMs. We propose a method called Align-KD to guide the student model to learn the cross-modal matching that occurs at the shallow layer. The teacher also helps student learn the projection of vision token into text embedding space based on the focus of text. Under the guidance of Align-KD, the 1.7B MobileVLM V2 model can learn rich knowledge from the 7B teacher model with light design of training loss, and achieve an average score improvement of 2.0 across 6 benchmarks under two training subsets respectively. Code is available at: https://github.com/fqhank/Align-KD.

Real-world image super-resolution (Real-ISR) aims to reconstruct high-resolution images from low-resolution inputs degraded by complex, unknown processes. While many Stable Diffusion (SD)-based Real-ISR methods have achieved remarkable success, their slow, multi-step inference hinders practical deployment. Recent SD-based one-step networks like OSEDiff and S3Diff alleviate this issue but still incur high computational costs due to their reliance on large pretrained SD models. This paper proposes a novel Real-ISR method, AdcSR, by distilling the one-step diffusion network OSEDiff into a streamlined diffusion-GAN model under our Adversarial Diffusion Compression (ADC) framework. We meticulously examine the modules of OSEDiff, categorizing them into two types: (1) Removable (VAE encoder, prompt extractor, text encoder, etc.) and (2) Prunable (denoising UNet and VAE decoder). Since direct removal and pruning can degrade the model's generation capability, we pretrain our pruned VAE decoder to restore its ability to decode images and employ adversarial distillation to compensate for performance loss. This ADC-based diffusion-GAN hybrid design effectively reduces complexity by 73% in inference time, 78% in computation, and 74% in parameters, while preserving the model's generation capability. Experiments manifest that our proposed AdcSR achieves competitive recovery quality on both synthetic and real-world datasets, offering up to 9.3x speedup over previous one-step diffusion-based methods. Code and models are available at https://github.com/Guaishou74851/AdcSR.

In this work, we introduce the Adventurer series models where we treat images as sequences of patch tokens and employ uni-directional language models to learn visual representations. This modeling paradigm allows us to process images in a recurrent formulation with linear complexity relative to the sequence length, which can effectively address the memory and computation explosion issues posed by high-resolution and fine-grained images. In detail, we introduce two simple designs that seamlessly integrate image inputs into the causal inference framework: a global pooling token placed at the beginning of the sequence and a flipping operation between every two layers. Extensive empirical studies highlight that compared with the existing plain architectures such as DeiT and Vim, Adventurer offers an optimal efficiency-accuracy trade-off. For example, our Adventurer-Base attains a competitive test accuracy of 84.3% on the standard ImageNet-1k benchmark with 216 images/s training throughput, which is 3.8x and 6.2x faster than Vim and DeiT to achieve the same result. As Adventurer offers great computation and memory efficiency and allows scaling with linear complexity, we hope this architecture can benefit future explorations in modeling long sequences for high-resolution or fine-grained images.

Semantic future prediction is important for autonomous systems navigating dynamic environments. This paper introduces FUTURIST, a method for multimodal future semantic prediction that uses a unified and efficient visual sequence transformer architecture. Our approach incorporates a multimodal masked visual modeling objective and a novel masking mechanism designed for multimodal training. This allows the model to effectively integrate visible information from various modalities, improving prediction accuracy. Additionally, we propose a VAE-free hierarchical tokenization process, which reduces computational complexity, streamlines the training pipeline, and enables end-to-end training with high-resolution, multimodal inputs. We validate FUTURIST on the Cityscapes dataset, demonstrating state-of-the-art performance in future semantic segmentation for both short- and mid-term forecasting. We provide the implementation code and model weights at https://github.com/Sta8is/FUTURIST.

In rapidly evolving field of vision-language models (VLMs), contrastive language-image pre-training (CLIP) has made significant strides, becoming foundation for various downstream tasks. However, relying on one-to-one (image, text) contrastive paradigm to learn alignment from large-scale messy web data, CLIP faces a serious myopic dilemma, resulting in biases towards monotonous short texts and shallow visual expressivity. To overcome these issues, this paper advances CLIP into one novel holistic paradigm, by updating both diverse data and alignment optimization. To obtain colorful data with low cost, we use image-to-text captioning to generate multi-texts for each image, from multiple perspectives, granularities, and hierarchies. Two gadgets are proposed to encourage textual diversity. To match such (image, multi-texts) pairs, we modify the CLIP image encoder into multi-branch, and propose multi-to-multi contrastive optimization for image-text part-to-part matching. As a result, diverse visual embeddings are learned for each image, bringing good interpretability and generalization. Extensive experiments and ablations across over ten benchmarks indicate that our holistic CLIP significantly outperforms existing myopic CLIP, including image-text retrieval, open-vocabulary classification, and dense visual tasks. Project page is available to further promote the prosperity of VLMs: https://voide1220.github.io/Holism/.

Sensory training during the early ages is vital for human development. Inspired by this cognitive phenomenon, we observe that the early training stage is also important for the multimodal learning process, where dataset information is rapidly acquired. We refer to this stage as the prime learning window. However, based on our observation, this prime learning window in multimodal learning is often dominated by information-sufficient modalities, which in turn suppresses the information acquisition of information-insufficient modalities. To address this issue, we propose Information Acquisition Regulation (InfoReg), a method designed to balance information acquisition among modalities. Specifically, InfoReg slows down the information acquisition process of information-sufficient modalities during the prime learning window, which could promote information acquisition of information-insufficient modalities. This regulation enables a more balanced learning process and improves the overall performance of the multimodal network. Experiments show that InfoReg outperforms related multimodal imbalanced methods across various datasets, achieving superior model performance. The code is available at https://github.com/GeWu-Lab/InfoReg_CVPR2025.

Vision-language models (VLMs) like CLIP have been widely used in various specific tasks.Parameter-efficient fine-tuning (PEFT) methods, such as prompt and adapter tuning,have become key techniques for adapting these models to specific domains.However, existing approaches rely on prior knowledgeto manually identify the locations requiring fine-tuning.Adaptively selecting which parameters in VLMs should be tuned remains unexplored. In this paper, we propose CLIP with Adaptive Selective Tuning (CLIP-AST), which can be used to automatically select critical parameters in VLMs for fine-tuning for specific tasks.It opportunely leveragesthe adaptive learning rate in the optimizer and improves model performance without extra parameter overhead. We conduct extensive experiments on 13 benchmarks, such as ImageNet, Food101, Flowers102, etc,with different settings, including few-shot learning, base-to-novel class generalization, and out-of-distribution. The results show that CLIP-AST consistently outperforms the original CLIP model as well as its variantsand achieves state-of-the-art (SOTA) performance in all cases. For example, with the 16-shot learning, CLIP-AST surpasses GraphAdapter and PromptSRC by 3.56% and 2.20% in average accuracy on 11 datasets, respectively.Code will be publicly available.

Multimodal large language models (MLLMs) have enabled open-world visual understanding by injecting visual input as extra tokens into large language models (LLMs) as contexts. However, when the visual input changes from a single image to a long video, the above paradigm encounters difficulty because the vast amount of video tokens has significantly exceeded the maximal capacity of MLLMs. Therefore, existing video-based MLLMs are mostly established upon sampling a small portion of tokens from input data, which can cause key information to be lost and thus produce incorrect answers. This paper presents a simple yet effective algorithm named Adaptive Keyframe Sampling (AKS). It inserts a plug-and-play module known as keyframe selection, which aims to maximize the useful information with a fixed number of video tokens. We formulate keyframe selection as an optimization involving (1) the relevance between the keyframes and the prompt, and (2) the coverage of the keyframes over the video, and present an adaptive algorithm to approximate the best solution. Experiments on two long video understanding benchmarks validate that AKS improves video QA accuracy (beyond strong baselines) upon selecting informative keyframes. Our study reveals the importance of information pre-filtering in video-based MLLMs. Our code are available at https://github.com/ncTimTang/AKS

Diffusion-based Text-to-Image (T2I) models have demonstrated significant potential in image restoration. However, existing models continue to grapple with challenges such as complex training and prompt design. We introduce a new perspective for improving image restoration by injecting knowledge from pretrained vision-language models into current T2I models. We empirically show that the degradation and content representations in BLIP-2 can be linearly separated, providing promising degradation guidance for image restoration. Specifically, the Feature Difference Instruction (FDI) is first extracted by Q-Formers through a simple subtraction operation based on reference image pairs. Then, we propose a multi-scale FDI adapter to decouple the degradation style and corrupted artifacts, and inject the styleflow exclusively into specific blocks through adapter-tuning, thereby preventing noise interference and eschewing the need for cumbersome weight retraining. In this way, we can train various task-specific adapters according to different degradations, achieving rich detail enhancement in the restoration results. Furthermore, the proposed FDI adapters have attractive properties of practical value, such as composability and generalization ability for all-in-one and mixed-degradation restoration. Extensive experiments under various settings demonstrate that our method has promising repairing quality over 10 image restoration tasks and a wide range of other applications.

Current deep homography estimation methods are typically constrained to processing low-resolution image pairs due to network architecture and computational limitations. For high-resolution images, downsampling is often required, which can greatly degrade estimation accuracy. In contrast, image matching methods, which match pixels and compute homography from correspondences, provide greater resolution flexibility. So in this work, we revisit the traditional image matching paradigm for homography estimation and propose GFNet, a Grid Flow regression Network that adapts the high-accuracy dense matching framework for homography estimation while enhancing efficiency through a grid-based strategy--estimating flow only over a coarse grid by leveraging homography's global smoothness. We demonstrate the effectiveness of GFNet on a wide range of experiments on multiple datasets, including the common scene MSCOCO, multimodal datasets VIS-IR and GoogleMap, and the dynamic scene VIRAT. Notably, on 448x448 GoogleMap, GFNet achieves an improvement of +13.5% in auc@3 while reducing MACs by ~47% compared to the SOTA dense matching method. Additionally, it shows a 1.8ximprovement in auc@3 over the SOTA deep homography method. Code is available at \textcolor[rgb] 0.95, 0.08, 0.58 https://github.com/KN-Zhang/GFNet .