CVPR 2025 论文列表

In this work, we introduce the first autoregressive framework for real-time, audio-driven portrait animation, a.k.a, talking head. Beyond the challenge of lengthy animation times, a critical challenge in realistic talking head generation lies in preserving the natural movement of diverse body parts. To this end, we propose Teller, the first streaming audio-driven protrait animation framework with autoregressive motion generation. Specifically, Teller first decomposes facial and body detail animation into two components: Facial Motion Latent Generation (FMLG) based on an autoregressive transfromer, and movement authenticity refinement using a Efficient Temporal Module (ETM).Concretely, FMLG employs a Residual VQ model to map the facial motion latent from the implicit keypoint-based model into discrete motion tokens, which are then temporally sliced with audio embeddings. This enables the AR tranformer to learn real-time, stream-based mappings from audio to motion.Furthermore, Teller incorporate ETM to capture finer motion details. This module ensures the physical consistency of body parts and accessories, such as neck muscles and earrings, improving the realism of these movements.Teller is designed to be efficient, surpassing the inference speed of diffusion-based models (Hallo 20.93s vs. Teller 0.92s for one second video generation), and achieves a real-time streaming performance of up to 25 FPS. Extensive experiments demonstrate that our method outperforms recent audio-driven portrait animation models, especially in small movements, as validated by human evaluations with a significant margin in quality and realism.

With the rapid advancement of Multi-modal Large Language Models (MLLMs), MLLM-based Image Quality Assessment (IQA) methods have shown promising performance in linguistic quality description. However, current methods still fall short in accurately scoring image quality. In this work, we aim to leverage MLLMs to regress accurate quality scores. A key challenge is that the quality score is inherently continuous, typically modeled as a Gaussian distribution, whereas MLLMs generate discrete token outputs. This mismatch necessitates score discretization. Previous approaches discretize the mean score into a one-hot label, resulting in information loss and failing to capture inter-image relationships. We propose a distribution-based approach that discretizes the score distribution into a soft label. This method preserves the characteristics of the score distribution, achieving high accuracy and maintaining inter-image relationships. Moreover, to address dataset variation, where different IQA datasets exhibit various distributions, we introduce a fidelity loss based on Thurstone's model. This loss captures intra-dataset relationships, facilitating co-training across multiple IQA datasets. With these designs, we develop the distribution-based Depicted image Quality Assessment model for Score regression (DeQA-Score). Experiments across multiple benchmarks show that DeQA-Score stably outperforms baselines in score regression. Also, DeQA-Score can predict the score distribution that closely aligns with human annotations. Codes and model weights have been released in https://depictqa.github.io/deqa-score/.

When a vision-language model (VLM) is prompted to identify an entity depicted in an image, it may answer "I see a conifer," rather than the specific label "Norway spruce". This raises two issues for evaluation: Firstly, the unconstrained generated text needs to be mapped to the evaluation label space (i.e., "conifer"). Secondly, a useful classification measure should give partial credit to less specific, but not incorrect, answers ("Norway spruce" being a type of "conifer"). To meet these requirements, we propose a framework for evaluating unconstrained text predictions such as those generated from a vision-language model against a taxonomy. Specifically, we propose the use of hierarchical precision and recall measures to assess the level of correctness and specificity of predictions with regard to a taxonomy. Experimentally, we first show that existing text similarity measures do not capture taxonomic similarity well. We then develop and compare different methods to map textual VLM predictions onto a taxonomy. This allows us to compute hierarchical similarity measures between the generated text and the ground truth labels. Finally, we analyze modern VLMs on fine-grained visual classification tasks based on our proposed taxonomic evaluation scheme.

Semi-supervised crowd counting is crucial for addressing the high annotation costs of densely populated scenes. Although several methods based on pseudo-labeling have been proposed, it remains challenging to effectively and accurately utilize unlabeled data. In this paper, we propose a novel framework called Taste More Taste Better (TMTB), which emphasizes both data and model aspects. Firstly, we explore a data augmentation technique well-suited for the crowd counting task. By inpainting the background regions, this technique can effectively enhance data diversity while preserving the fidelity of the entire scenes. Secondly, we introduce the Visual State Space Model as backbone to capture the global context information from crowd scenes, which is crucial for extremely crowded, low-light, and adverse weather scenarios. In addition to the traditional regression head for exact prediction, we employ an Anti-Noise classification head to provide less exact but more accurate supervision, since the regression head is sensitive to noise in manual annotations. We conduct extensive experiments on four benchmark datasets and show that our method outperforms state-of-the-art methods by a large margin. Code is publicly available on https://github.com/syhien/taste_more_taste_better.

Multi-modal image fusion aggregates information from multiple sensor sources, achieving superior visual quality and perceptual features compared to single-source images, often improving downstream tasks. However, current fusion methods for downstream tasks still use predefined fusion objectives that potentially mismatch the downstream tasks, limiting adaptive guidance and reducing model flexibility. To address this, we propose Task-driven Image Fusion (TDFusion), a fusion framework incorporating a learnable fusion loss guided by task loss. Specifically, our fusion loss includes learnable parameters modeled by a neural network called the loss generation module. This module is supervised by the downstream task loss in a meta-learning manner. The learning objective is to minimize the task loss of fused images after optimizing the fusion module with the fusion loss. Iterative updates between the fusion module and the loss module ensure that the fusion network evolves toward minimizing task loss, guiding the fusion process toward the task objectives. TDFusion's training relies entirely on the downstream task loss, making it adaptable to any specific task. It can be applied to any architecture of fusion and task networks. Experiments demonstrate TDFusion's performance through fusion experiments conducted on four different datasets, in addition to evaluations on semantic segmentation and object detection tasks. The code is available at https://github.com/HaowenBai/TDFusion.

The goal of visual grounding is to establish connections between target objects and textual descriptions. Large Language Models (LLMs) have demonstrated strong comprehension abilities across a variety of visual tasks. To establish precise associations between the text and the corresponding visual region, we propose a Task-aware Cross-modal feature Refinement Transformer with LLMs for visual grounding, and our model is referred to as TCRT. To enable the LLM trained solely on text to understand images, we introduce an LLM adaptation module that extracts text-related visual features to bridge the domain discrepancy between the textual and visual modalities. We feed the text and visual features into the LLM to obtain task-aware priors. To enable the priors to guide feature fusion process, we further incorporate a cross-modal feature fusion module, which allows task-aware embeddings to refine visual features and facilitate information interaction between the Referring Expression Comprehension (REC) and Referring Expression Segmentation (RES) tasks. We have performed extensive experiments to verify the effectiveness of the main components and demonstrate the superior performance of the proposed TCRT over state-of-the-art end-to-end visual grounding methods on RefCOCO, RefCOCOg, RefCOCO+ and ReferItGame.

Optimization-based meta-learning methods for few-shot one-class classification (FS-OCC) aim to fine-tune a meta-trained model to classify the positive and negative samples using only a few positive samples by adaptation. However, recent approaches primarily focus on adjusting existing meta-learning algorithms for FS-OCC, while overlooking issues stemming from the misalignment between the cross-entropy loss and OCC tasks during adaptation. This misalignment, combined with the limited availability of one-class samples and the restricted diversity of task-specific adaptation, can significantly exacerbate the adverse effects of gradient instability and generalization. To address these challenges, we propose a novel Task-Specific Gradient Adaptation (TSGA) for FS-OCC. Without extra supervision, TSGA learns to generate appropriate, stable gradients by leveraging label prediction and feature representation details of one-class samples and refines the adaptation process by recalibrating task-specific gradients and regularization terms. We evaluate TSGA on three challenging datasets and a real-world CNC Milling Machine application and demonstrate consistent improvements over baseline methods. Furthermore, we illustrate the critical impact of gradient instability and task-agnostic adaptation. Notably, TSGA achieves state-of-the-art results by effectively addressing these issues.

Prompt learning has attracted widespread attention in adapting vision-language models to downstream tasks. Existing methods largely rely on optimization strategies to ensure the task-awareness of learnable prompts. Due to the scarcity of task-specific data, overfitting is prone to occur. The resulting prompts often do not generalize well or exhibit limited task-awareness. To address this issue, we propose a novel Task-Aware Clustering (TAC) framework for prompting vision-language models, which increases the task-awareness of learnable prompts by introducing task-aware pre-context. The key ingredients are as follows: (a) generating task-aware pre-context based on task-aware clustering that can preserve the backbone structure of a downstream task with only a few clustering centers, (b) enhancing the task-awareness of learnable prompts by enabling them to interact with task-aware pre-context via the well-pretrained encoders, and (c) preventing the visual task-aware pre-context from interfering the interaction between patch embeddings by masked attention mechanism. Extensive experiments are conducted on benchmark datasets, covering the base-to-novel, domain generalization, and cross-dataset transfer settings. Ablation studies validate the effectiveness of key ingredients. Comparative results show the superiority of our TAC over competitive counterparts. The code is available at https://github.com/FushengHao/TAC.

The ability to learn new concepts while preserve the learned knowledge is desirable for learning systems in Class-Incremental Learning (CIL). Recently, feature expansion of the model become a prevalent solution for CIL, where the old features are fixed during the training of the new task while new features are expanded for the new tasks. However, such task-specific features learned from the new task may collide with the old features, leading to misclassification between tasks. Therefore, the expanded model is often encouraged to capture diverse features from the new task, aiming to avoid such collision. However, the existing solution is largely restricted to the samples from the current task, because of the poor accessibility to previous samples. To promote the learning and transferring of diverse features across tasks, we propose a framework called Task-Agnostic Guided Feature Expansion (TagFex). Firstly, it captures task-agnostic features continually with a separate model, providing extra task-agnostic features for subsequent tasks. Secondly, to obtain useful features from the task-agnostic model for the current task, it aggregates the task-agnostic features with the task-specific feature using a merge attention. Then the aggregated feature is transferred back into the task-specific feature for inference, helping the task-specific model capture diverse features. Extensive experiments show the effectiveness and superiority of TagFex on various CIL settings.

Task Arithmetic has emerged as a simple yet effective method to merge models without additional training. However, by treating entire networks as flat parameter vectors, it overlooks key structural information and is susceptible to task interference. In this paper, we study task vectors at the layer level, focusing on task layer matrices and their singular value decomposition. In particular, we concentrate on the resulting singular vectors, which we refer to as Task Singular Vectors (TSV). Recognizing that layer task matrices are often low-rank, we propose TSV-Compress, a simple procedure that compresses them to 10% of their original size while retaining 99% of accuracy. We further leverage this low-rank space to define a new measure of task interference based on the interaction of singular vectors from different tasks. Building on these findings, we introduce TSV-Merge, a novel model merging approach that combines compression with interference reduction, significantly outperforming existing methods.

Current multimodal large language models (MLLMs) struggle with fine-grained or precise understanding of visuals although they give comprehensive perception and reasoning in a spectrum of vision applications. Recent studies either develop tool-using or unify specific visual tasks into the autoregressive framework, often at the expense of overall multimodal performance. To address this issue and enhance MLLMs with visual tasks in a scalable fashion, we propose Task Preference Optimization (TPO), a novel method that utilizes differentiable task preferences derived from typical fine-grained visual tasks. TPO introduces learnable task tokens that establish connections between multiple task-specific heads and the MLLM. By leveraging rich visual labels during training, TPO significantly enhances the MLLM's multimodal capabilities and task-specific performance. Through multi-task co-training within TPO, we observe synergistic benefits that elevate individual task performance beyond what is achievable through single-task training methodologies. Our instantiation of this approach with VideoChat and LLaVA demonstrates an overall 14.6% improvement in multimodal performance compared to baseline models. Additionally, MLLM-TPO demonstrates robust zero-shot capabilities across various tasks, performing comparably to state-of-the-art supervised models.

Despite recent advances in deep learning, most existing learning IMU odometry methods are trained on specific datasets, lack generalization, and are prone to overfitting, which limits their real-world application. To address these challenges, we present Tartan IMU, a foundation model designed for generalizable, IMU-based state estimation across diverse robotic platforms. Our approach consists of three-stage: First, a pre-trained foundation model leverages over 100 hours of multi-platform data to establish general motion knowledge, achieving 36% improvement in ATE over specialized models. Second, to adapt to previously unseen tasks, we employ the Low-Rank Adaptation (LoRA), allowing positive transfer with only 1.1 M trainable parameters. Finally, to support robotics deployment, we introduce online test-time adaptation, which eliminates the boundary between training and testing, allowing the model to continuously "learn as it operates" at 200 FPS in real-time.

Foundation models (FMs) such as CLIP have demonstrated impressive zero-shot performance across various tasks by leveraging large-scale, unsupervised pre-training. However, they often inherit harmful or unwanted knowledge from noisy internet-sourced datasets, compromising their reliability in real-world applications. Existing model unlearning methods either rely on access to pre-trained datasets or focus on coarse-grained unlearning (e.g., entire classes), leaving a critical gap for fine-grained unlearning. In this paper, we address the challenging scenario of selectively forgetting specific portions of knowledge within a class--without access to pre-trained data--while preserving the model's overall performance. We propose a novel three-stage approach that progressively unlearns targeted knowledge while mitigating over-forgetting. It consists of (1) a forgetting stage to fine-tune the CLIP on samples to be forgotten, (2) a reminding stage to restore performance on retained samples, and (3) a restoring stage to recover zero-shot capabilities using model souping. Additionally, we introduce knowledge distillation to handle the distribution disparity between forgetting/retaining samples and unseen pre-trained data. Extensive experiments on CIFAR-10, ImageNet-1K, and style datasets demonstrate that our approach effectively unlearns specific subgroups while maintaining strong zero-shot performance on semantically similar subgroups and other categories, significantly outperforming baseline unlearning methods, which lose effectiveness under the CLIP unlearning setting.

Realistic 3D full-body talking avatars hold great potential in AR, with applications ranging from e-commerce live streaming to holographic communication. Despite advances in 3D Gaussian Splatting (3DGS) for lifelike avatar creation, existing methods struggle with fine-grained control of facial expressions and body movements in full-body talking tasks. Additionally, they often lack sufficient details and cannot run in real-time on mobile devices. We present TaoAvatar, a high-fidelity, lightweight, 3DGS-based full-body talking avatar driven by various signals. Our approach starts by creating a personalized clothed human parametric template that binds Gaussians to represent appearances. We then pre-train a StyleUnet-based network to handle complex pose-dependent non-rigid deformation, which can capture high-frequency appearance details but is too resource-intensive for mobile devices. To overcome this, we "bake" the non-rigid deformations into a lightweight MLP-based network using a distillation technique and develop blend shapes to compensate for details. Extensive experiments show that TaoAvatar achieves state-of-the-art rendering quality while running in real-time across various devices, maintaining 90 FPS on high-definition stereo devices such as the Apple Vision Pro.

Despite recent successes in novel view synthesis using 3D Gaussian Splatting (3DGS), modeling scenes with sparse inputs remains a challenge. In this work, we address two critical yet overlooked issues in real-world sparse-input modeling: extrapolation and occlusion. To tackle these issues, we propose to use a reconstruction by generation pipeline that leverages learned priors from video diffusion models to provide plausible interpretations for regions outside the field of view or occluded. However, the generated sequences exhibit inconsistencies that do not fully benefit subsequent 3DGS modeling. To address the challenge of inconsistency, we introduce a novel scene-grounding guidance based on rendered sequences from an optimized 3DGS, which tames the diffusion model to generate consistent sequences. This guidance is training-free and does not require any fine-tuning of the diffusion model. To facilitate holistic scene modeling, we also propose a trajectory initialization method. It effectively identifies regions that are outside the field of view and occluded. We further design a scheme tailored for 3DGS optimization with generated sequences. Experiments demonstrate that our method significantly improves upon the baseline and achieves state-of-the-art performance on challenging benchmarks.

We introduce MAGI, a hybrid video generation framework that combines masked modeling for intra-frame generation with causal modeling for next-frame generation. Our key innovation, Complete Teacher Forcing (CTF), conditions masked frames on complete observation frames rather than masked ones (namely Masked Teacher Forcing, MTF), enabling a smooth transition from token-level (patch-level) to frame-level autoregressive generation. CTF significantly outperforms MTF, achieving a 23% improvement in FVD scores on first-frame conditioned video prediction. To address issues like exposure bias, we employ targeted training strategies, setting a new benchmark in autoregressive video generation. Experiments show that MAGI can generate long, coherent video sequences exceeding 100 frames, even when trained on as few as 16 frames, highlighting its potential for scalable, high-quality video generation.

We aim to solve unsupervised anomaly detection in a practical challenging environment where the normal dataset is both contaminated with defective regions and its product class distribution is tailed but unknown. We observe that existing models suffer from tail-versus-noise trade-off where if a model is robust against pixel noise, then its performance deteriorates on tail class samples, and vice versa. To mitigate the issue, we handle the tail class and noise samples independently. To this end, we propose TailSampler, a novel class size predictor that estimates the class cardinality of samples based on a symmetric assumption on the class-wise distribution of embedding similarities. TailSampler can be utilized to sample the tail class samples exclusively, allowing to handle them separately. Based on these facets, we build a memory-based anomaly detection model TailedCore, whose memory both well captures tail class information and is noise-robust. We extensively validate the effectiveness of TailedCore on the unsupervised long-tail noisy anomaly detection setting, and show that TailedCore outperforms the state-of-the-art in most settings. Code is available in TailedCore

Radar-Camera depth estimation aims to predict dense and accurate metric depth by fusing input images and Radar data. Model efficiency is crucial for this task in pursuit of real-time processing on autonomous vehicles and robotic platforms. However, due to the sparsity of Radar returns, the prevailing methods adopt multi-stage frameworks with intermediate quasi-dense depth, which are time-consuming and not robust. To address these challenges, we propose TacoDepth, an efficient and accurate Radar-Camera depth estimation model with one-stage fusion. Specifically, the graph-based Radar structure extractor and the pyramid-based Radar fusion module are designed to capture and integrate the graph structures of Radar point clouds, delivering superior model efficiency and robustness without relying on the intermediate depth results. Moreover, TacoDepth can be flexible for different inference modes, providing a better balance of speed and accuracy. Extensive experiments are conducted to demonstrate the efficacy of our method. Compared with the previous state-of-the-art approach, TacoDepth improves depth accuracy and processing speed by 12.8% and 91.8%. Our work provides a new perspective on efficient Radar-Camera depth estimation.