CVPR 2025 Papers

Diffusion-based audio-driven talking avatar methods have recently gained attention for their high-fidelity, vivid, and expressive results. However, their slow inference speed limits practical applications. Despite the development of various distillation techniques for diffusion models, we found that naive diffusion distillation methods do not yield satisfactory results. Distilled models exhibit reduced robustness with open-set input images and a decreased correlation between audio and video compared to teacher models, undermining the advantages of diffusion models. To address this, we propose FADA (Fast Diffusion Avatar Synthesis with Mixed-Supervised Multi-CFG Distillation). We first designed a mixed-supervised loss to leverage data of varying quality and enhance the overall model capability as well as robustness. Additionally, we propose a multi-CFG distillation with learnable tokens to utilize the correlation between audio and reference image conditions, reducing the threefold inference runs caused by multi-CFG with acceptable quality degradation. Extensive experiments across multiple datasets show that FADA generates vivid videos comparable to recent diffusion model-based methods while achieving an NFE speedup of 4.17-12.5 times. Demos are available at our webpage https://fadavatar.github.io.

Endowing Large Multimodal Models (LMMs) with visual grounding capability can significantly enhance AIs' understanding of the visual world and their interaction with humans. However, existing methods typically fine-tune the parameters of LMMs to learn additional segmentation tokens and overfit grounding and segmentation datasets. Such a design would inevitably cause a catastrophic diminution in the indispensable conversational capability of general AI assistants. In this paper, we comprehensively evaluate state-of-the-art grounding LMMs across a suite of multimodal question-answering benchmarks, observing drastic performance drops that indicate vanishing general knowledge comprehension and weakened instruction following ability. To address this issue, we present F-LMM---grounding frozen off-the-shelf LMMs in human-AI conversations---a straightforward yet effective design based on the fact that word-pixel correspondences conducive to visual grounding inherently exist in the attention mechanism of well-trained LMMs. Using only a few trainable CNN layers, we can translate word-pixel attention weights to mask logits, which a SAM-based mask refiner can further optimise. Our F-LMM neither learns special segmentation tokens nor utilises high-quality grounded instruction-tuning data, but achieves competitive performance on referring expression segmentation and panoptic narrative grounding benchmarks while completely preserving LMMs' original conversational ability. Additionally, with instruction-following ability preserved and grounding ability obtained, F-LMM can be directly applied to complex tasks like reasoning segmentation, grounded conversation generation and visual chain-of-thought reasoning. Our code can be found at https://github.com/wusize/F-LMM.

We present a technique and benchmark dataset for estimating the relative 3D orientation between a pair of Internet images captured in an extreme setting, where the images have limited or non-overlapping field of views. Prior work targeting extreme rotation estimation assume constrained 3D environments and emulate perspective images by cropping regions from panoramic views. However, real images captured in the wild are highly diverse, exhibiting variation in both appearance and camera intrinsics. In this work, we propose a Transformer-based method for estimating relative rotations in extreme real-world settings, and contribute the ExtremeLandmarkPairs dataset, assembled from scene-level Internet photo collections. Our evaluation demonstrates that our approach succeeds in estimating the relative rotations in a wide variety of extreme-view Internet image pairs, outperforming various baselines, including dedicated rotation estimation techniques and contemporary 3D reconstruction methods.

Image exposure correction enhances images captured under diverse real-world conditions by addressing issues of under- and over-exposure, which can result in the loss of critical details and hinder content recognition. While significant advancements have been made, current methods often fail to achieve optimal feature learning for effective correction.To overcome these challenges, we propose Exposure-slot, a novel framework that integrates a prompt-based slot-in-slot attention mechanism to cluster exposed feature regions and learn exposure-centric features for each cluster. By extending the Slot Attention algorithm with a hierarchical structure, our approach progressively clusters features, enabling precise and region-aware correction. In particular, learnable prompts tailored to exposure characteristics of slots further enhance feature quality, adapting dynamically to varying conditions. Our method delivers superior performance on benchmark datasets, surpassing the current state-of-the-art with a PSNR improvement of over 1.85 dB on the SICE dataset and 0.4 dB on the LCDP dataset, thereby establishing a new benchmark for multi-exposure correction. The source code can be found at:https://github.com/kdhRick2222/Exposure-slot.

This paper does not describe a new method; instead, it provides a thorough exploration of an important yet understudied design space related to recent advances in text-to-image synthesis---specifically, the deep fusion of large language models (LLMs) with diffusion transformers (DiTs) for multimodal generation. Previous studies mainly focused on overall system performance rather than detailed comparisons with alternative methods, and key design details and training recipes were often left undisclosed. These gaps create uncertainty about the real potential of this approach. To fill these gaps, we conduct an empirical study on text-to-image generation, performing controlled comparisons with established baselines, analyzing important design choices, and providing a clear, reproducible recipe for training at scale. We hope this work offers meaningful data points and practical guidelines for future research in multimodal generation.

Despite inheriting security measures from underlying language models, Vision-Language Models (VLMs) may still be vulnerable to safety alignment issues. Through empirical analysis, we uncover two critical findings: scenario-matched images can significantly amplify harmful outputs, and contrary to common assumptions in gradient-based attacks, minimal loss values do not guarantee optimal attack effectiveness. Building on these insights, we introduce MLAI (Multi-Loss Adversarial Images), a novel jailbreak framework that leverages scenario-aware image generation for semantic alignment, exploits flat minima theory for robust adversarial image selection, and employs multi-image collaborative attacks for enhanced effectiveness. Extensive experiments demonstrate MLAI's significant impact, achieving attack success rates of 77.75% on MiniGPT-4 and 82.80% on LLaVA-2, substantially outperforming existing methods by margins of 34.37% and 12.77% respectively. Furthermore, MLAI shows considerable transferability to commercial black-box VLMs, achieving up to 60.11% success rate. Our work reveals fundamental visual vulnerabilities in current VLMs safety mechanisms and underscores the need for stronger defenses. Warning: This paper contains potentially harmful example text.

This paper introduces a new control signal for facial motion generation: timeline control. Compared to audio and text signals, timelines provide more fine-grained control, such as generating specific facial motions with precise timing. Users can specify a multi-track timeline of facial actions arranged in temporal intervals, allowing precise control over the timing of each action. To model the timeline control capability, We first annotate the time intervals of facial actions in natural facial motion sequences at a frame-level granularity. This process is facilitated by Toeplitz Inverse Covariance-based Clustering to minimize human labor. Based on the annotations, we propose a diffusion-based generation model capable of generating facial motions that are natural and accurately aligned with input timelines. Our method supports text-guided motion generation by using ChatGPT to convert text into timelines. Experimental results show that our method can annotate facial action intervals with satisfactory accuracy, and produces natural facial motions accurately aligned with timelines.

Point tracking in videos is a fundamental task with applications in robotics, video editing, and more. While many vision tasks benefit from pre-trained feature backbones to improve generalizability, point tracking has primarily relied on simpler backbones trained from scratch on synthetic data, which may limit robustness in real-world scenarios. Additionally, point tracking requires temporal awareness to ensure coherence across frames, but using temporally-aware features is still underexplored. Most current methods often employ a two-stage process: an initial coarse prediction followed by a refinement stage to inject temporal information and correct errors from the coarse stage. These approach, however, is computationally expensive and potentially redundant if the feature backbone itself captures sufficient temporal information. In this work, we introduce Chrono, a feature backbone specifically designed for point tracking with built-in temporal awareness. Leveraging pre-trained representations from self-supervised learner DINOv2 and enhanced with a temporal adapter, Chrono effectively captures long-term temporal context, enabling precise prediction even without the refinement stage. Experimental results demonstrate that Chrono achieves state-of-the-art performance in a refiner-free setting on the TAP-Vid-DAVIS and TAP-Vid-Kinetics datasets, among common feature backbones used in point tracking as well as DINOv2, with exceptional efficiency.

Due to the difficulty in scaling up, generative adversarial networks (GANs) seem to be falling out of grace with the task of text-conditioned image synthesis. Sparsely activated mixture-of-experts (MoE) has recently been demonstrated as a valid solution to training large-scale models with limited resources. Inspired by this, we present Aurora, a GAN-based text-to-image generator that employs a collection of experts to learn feature processing, together with a sparse router to adaptively select the most suitable expert for each feature point. We adopt a two-stage training strategy, which first learns a base model at 64x64 resolution followed by an upsampler to produce 512x512 images. Trained with only public data, our approach encouragingly closes the performance gap between GANs and industry-level diffusion models, maintaining a fast inference speed. We release the code and checkpoints \href https://github.com/zhujiapeng/Aurora here to facilitate the community for further development.

Open-vocabulary semantic segmentation models aim to accurately assign a semantic label to each pixel in an image from a set of arbitrary open-vocabulary texts. In order to learn such pixel-level alignment, current approaches typically rely on a combination of (i) image-level VL model (e.g. CLIP), (ii) ground truth masks, (iii) custom grouping encoders, and (iv) the Segment Anything Model (SAM). In this paper, we introduce S-Seg, a simple model that can achieve surprisingly strong performance without depending on any of the above elements. S-Seg leverages pseudo-mask and language to train a MaskFormer, and can be easily trained from publicly available image-text datasets. Contrary to prior works, our model directly trains for pixel-level features and language alignment. Once trained, S-Seg generalizes well to multiple testing datasets without requiring fine-tuning. In addition, S-Seg has the extra benefits of scalability with data and consistently improving when augmented with self-training. We believe that our simple yet effective approach will serve as a solid baseline for future research. Our code and demo will be made publicly available soon.

Generative Adversarial Networks (GANs) have been widely applied to image super-resolution (SR) to enhance the perceptual quality. However, most existing GAN-based SR methods typically perform coarse-grained discrimination directly on images and ignore the semantic information of images, making it challenging for the super resolution networks (SRN) to learn fine-grained and semantic-related texture details. To alleviate this issue, we propose a semantic feature discrimination method, SFD, for perceptual SR. Specifically, we first design a feature discriminator (Feat-D), to discriminate the pixel-wise middle semantic features from CLIP, aligning the feature distributions of SR images with that of high-quality images. Additionally, we propose a text-guided discrimination method (TG-D) by introducing learnable prompt pairs (LPP) in an adversarial manner to perform discrimination on the more abstract output feature of CLIP, further enhancing the discriminative ability of our method. With both Feat-D and TG-D, our SFD can effectively distinguish between the semantic feature distributions of low-quality and high-quality images, encouraging SRN to generate more realistic and semantic-relevant textures. Furthermore, based on the trained Feat-D and LPP, we propose a novel opinion-unaware no-reference image quality assessment (OU NR-IQA) method, SFD-IQA, greatly improving OU NR-IQA performance without any additional targeted training. Extensive experiments on classical SISR, real-world SISR, and OU NR-IQA tasks, demonstrate the effectiveness of our proposed methods. Code is available at https://github.com/GuangluDong0728/SFD.

This paper explores scene affinity (AIScene), namely intra-scene consistency and inter-scene correlation, for semi-supervised LiDAR semantic segmentation in driving scenes. Adopting teacher-student training, AIScene employs a teacher network to generate pseudo-labeled scenes from unlabeled data, which then supervise the student network's learning. Unlike most methods that include all points in pseudo-labeled scenes for forward propagation but only pseudo-labeled points for backpropagation, AIScene removes points without pseudo-labels, ensuring consistency in both forward and backward propagation within the scene. This simple point erasure strategy effectively prevents unsupervised, semantically ambiguous points (excluded in backpropagation) from affecting the learning of pseudo-labeled points. Moreover, AIScene incorporates patch-based data augmentation, mixing multiple scenes at both scene and instance levels. Compared to existing augmentation techniques that typically perform scene-level mixing between two scenes, our method enhances the semantic diversity of labeled (or pseudo-labeled) scenes, thereby improving the semi-supervised performance of segmentation models. Experiments show that AIScene outperforms previous methods on two popular benchmarks across four settings, achieving notable improvements of 1.9% and 2.1% in the most challenging 1% labeled data. The code will be released at https://github.com/azhuantou/AIScene.

Anomaly detection (AD) is essential for industrial inspection, yet existing methods typically rely on "comparing" test images to normal references from a training set. However, variations in appearance and positioning often complicate the alignment of these references with the test image, limiting detection accuracy. We observe that most anomalies manifest as local variations, meaning that even within anomalous images, valuable normal information remains. We argue that this information is useful and may be more aligned with the anomalies since both the anomalies and the normal information originate from the same image. Therefore, rather than relying on external normality from the training set, we propose INP-Former, a novel method that extracts Intrinsic Normal Prototypes (INPs) directly from the test image. Specifically, we introduce the INP Extractor, which linearly combines normal tokens to represent INPs. We further propose an INP Coherence Loss to ensure INPs can faithfully represent normality for the testing image. These INPs then guide the INP-Guided Decoder to reconstruct only normal tokens, with reconstruction errors serving as anomaly scores. Additionally, we propose a Soft Mining Loss to prioritize hard-to-optimize samples during training. INP-Former achieves state-of-the-art performance in single-class, multi-class, and few-shot AD tasks across MVTec-AD, VisA, and Real-IAD, positioning it as a versatile and universal solution for AD. Remarkably, INP-Former also demonstrates some zero-shot AD capability. Code is available at: https://github.com/luow23/INP-Former.

Combining the advantages of conventional and event cameras for robust visual tracing has drawn extensive interest. However, existing tracking approaches heavily engage in complex cross-modal fusion modules, leading to higher computational complexity and training challenges. Besides, these methods generally ignore the effective integration of historical information, which is crucial to grasping the change in the target's appearance and motion trends. Given the recent advancements in Mamba's long-range modeling and linear complexity, we explore its potential in addressing the above issues in RGBE tracking tasks. Specifically, we first propose an efficient fusion module based on Mamba, which utilizes a simple gate-based interaction scheme to achieve effective modality-selective fusion. This module can be seamlessly integrated into the encoding layer of prevalent Transformer-based backbones. Moreover, we further present a novel historical decoder that leverages Mamba's advanced long sequence modeling to effectively capture the target appearance changes with autoregressive queries. Extensive experiments show that our proposed approach achieves state-of-the-art performance on multiple challenging short-term and long-term RGBE benchmarks. Besides, the effectiveness of each key Mamba-based component of our approach is evidenced by our thorough ablation study.

Referring expression counting (REC) algorithms are for more flexible and interactive counting ability across varied fine-grained text expressions. However, the requirement for fine-grained attribute understanding poses challenges for prior arts, as they struggle to accurately align attribute information with correct visual patterns. Given the proven importance of "visual density", it is presumed that the limitations of current REC approaches stem from an under-exploration of "contextual attribute density" (CAD). In the scope of REC, we define the CAD as the measure of the information intensity of one certain fine-grained attribute in visual regions. To model the the CAD, we propose a U-shape CAD estimator in which referring expression and multi-scale visual features from GroundingDINO can interact with each other. With additional density supervisions, we can effectively encode CAD, which is subsequently decoded via a novel attention procedure with CAD-refined queries. Integrating all these contributions, our framework significantly outperforms state-of-the-art REC methods, achieves 30% error reduction in counting metics and a 10% improvement in localization accuracy. The surprising results shed lights on the significance of contextual attribute density for REC. Code will be at github.com/Xu3XiWang/CAD-GD.

Weakly Supervised Semantic Segmentation (WSSS) with image-level labels aims to achieve pixel-level predictions using Class Activation Maps (CAMs). Recently, Contrastive Language-Image Pre-training (CLIP) has been introduced in WSSS. However, recent methods primarily focus on image-text alignment for CAM generation, while CLIP's potential in patch-text alignment remains unexplored. In this work, we propose ExCEL to explore CLIP's dense knowledge via a novel patch-text alignment paradigm for WSSS. Specifically, we propose Text Semantic Enrichment (TSE) and Visual Calibration (VC) modules to improve the dense alignment across both text and vision modalities. To make text embeddings semantically informative, our TSE module applies Large Language Models (LLMs) to build a dataset-wide knowledge base and enriches the text representations with an implicit attribute-hunting process. To mine fine-grained knowledge from visual features, our VC module first proposes Static Visual Calibration (SVC) to propagate fine-grained knowledge in a non-parametric manner. Then Learnable Visual Calibration (LVC) is further proposed to dynamically shift the frozen features towards distributions with diverse semantics. With these enhancements, ExCEL not only retains CLIP's training-free advantages but also significantly outperforms other state-of-the-art methods with much less training cost on PASCAL VOC and MS COCO. Code is available at https://github.com/zwyang6/ExCEL.

Modern world models require costly and time-consuming collection of large video datasets with action demonstrations by people or by environment-specific agents. To simplify training, we focus on using many virtual environments for inexpensive, automatically collected interaction data. Genie, a recent multi-environment world model, demonstrates simulation abilities of many environments with shared behavior. Unfortunately, training their model requires expensive demonstrations. Therefore, we propose a training framework merely using a random agent in virtual environments. While the model trained in this manner exhibits good controls, it is limited by the random exploration possibilities. To address this limitation, we propose AutoExplore Agent - an exploration agent that entirely relies on the uncertainty of the world model, delivering diverse data from which it can learn the best. Our agent is fully independent of environment-specific rewards and thus adapts easily to new environments. With this approach, the pretrained multi-environment model can quickly adapt to new environments achieving video fidelity and controllability improvement. In order to obtain automatically large-scale interaction datasets for pretraining, we group environments with similar behavior and controls. To this end, we annotate the behavior and controls of 974 virtual environments - a dataset that we name RetroAct. For building our model, we first create an open implementation of Genie - GenieRedux and apply enhancements and adaptations in our version GenieRedux-G. Our code and data are available at https://github.com/insait-institute/GenieRedux

Video semantic segmentation (VSS) plays a vital role in understanding the temporal evolution of scenes. Traditional methods often segment videos frame-by-frame or in a short temporal window, leading to limited temporal context, redundant computations, and heavy memory requirements. To this end, we introduce a Temporal Video State Space Sharing (TV3S) architecture to leverage Mamba state space models for temporal feature sharing. Our model features a selective gating mechanism that efficiently propagates relevant information across video frames, eliminating the need for a memory-heavy feature pool. By processing spatial patches independently and incorporating shifted operation, TV3S supports highly parallel computation in both training and inference stages, which reduces the delay in sequential state space processing and improves the scalability for long video sequences. Moreover, TV3S incorporates information from prior frames during inference, achieving long-range temporal coherence and superior adaptability to extended sequences. Evaluations on the VSPW and Cityscapes datasets reveal that our approach outperforms current state-of-the-art methods, establishing a new standard for VSS with consistent results across long video sequences. By achieving a good balance between accuracy and efficiency, TV3S shows a significant advancement in spatiotemporal modeling, paving the way for efficient video analysis. The code is publicly available at https://github.com/Ashesham/TV3S.git.

In this paper, we propose DeepDeblurRF, a novel radiance field deblurring approach that can synthesize high-quality novel views from blurred training views with significantly reduced training time. DeepDeblurRF leverages deep neural network (DNN)-based deblurring modules to enjoy their deblurring performance and computational efficiency. To effectively combine DNN-based deblurring and radiance field construction, we propose a novel radiance field (RF)-guided deblurring and an iterative framework that performs RF-guided deblurring and radiance field construction in an alternating manner. Moreover, DeepDeblurRF is compatible with various scene representations, such as voxel grids and 3D Gaussians, expanding its applicability. We also present BlurRF-Synth, the first large-scale synthetic dataset for training radiance field deblurring frameworks. We conduct extensive experiments on both camera motion blur and defocus blur, demonstrating that DeepDeblurRF achieves state-of-the-art novel-view synthesis quality with significantly reduced training time.