CVPR 2025 Papers

The relationship between muscle activity and resulting facial expressions is crucial for various fields, including psychology, medicine, and entertainment. The synchronous recording of facial mimicry and muscular activity via surface electromyography (sEMG) provides a unique window into these complex dynamics. Unfortunately, existing methods for facial analysis cannot handle electrode occlusion, rendering them ineffective. Even with occlusion-free reference images of the same person, variations in expression intensity and execution are unmatchable. Our electromyography-informed facial expression reconstruction (EIFER) approach is a novel method to restore faces under sEMG occlusion faithfully in an adversarial manner. We decouple facial geometry and visual appearance (e.g., skin texture, lighting, electrodes) by combining a 3D Morphable Model (3DMM) with neural unpaired image-to-image translation via reference recordings. Then, EIFER learns a bidirectional mapping between 3DMM expression parameters and muscle activity, establishing correspondence between the two domains. We validate the effectiveness of our approach through experiments on a dataset of synchronized sEMG recordings and facial mimicry, demonstrating faithful geometry and appearance reconstruction. Further, we synthesize expressions based on muscle activity and how observed expressions can predict dynamic muscle activity. Consequently, EIFER introduces a new paradigm for facial electromyography, which could be extended to other forms of multi-modal face recordings.

Principal Component Analysis (PCA), a classical dimensionality reduction technique, and 2D Gaussian representation, an adaptation of 3D Gaussian Splatting for image representation, offer distinct approaches to modeling visual data. We present EigenGS, a novel method that bridges these paradigms through an efficient transformation pipeline connecting eigenspace and image-space Gaussian representations. Our approach enables instant initialization of Gaussian parameters for new images without requiring per-image optimization from scratch, dramatically accelerating convergence. EigenGS introduces a frequency-aware learning mechanism that encourages Gaussians to adapt to different scales, effectively modeling varied spatial frequencies and preventing artifacts in high-resolution reconstruction. Extensive experiments demonstrate that EigenGS not only achieves superior reconstruction quality compared to direct 2D Gaussian fitting but also reduces the necessary parameter count and training time. The results highlight EigenGS's effectiveness and generalization ability across images with varying resolutions and diverse categories, making Gaussian-based image representation both high-quality and viable for real-time applications.

We introduce EgoTextVQA, a novel and rigorously constructed benchmark for egocentric QA assistance involving scene text. EgoTextVQA contains 1.5K ego-view videos and 7K scene-text aware questions that reflect real user needs in outdoor driving and indoor house-keeping activities. The questions are designed to elicit identification and reasoning on scene text in an egocentric and dynamic environment. With EgoTextVQA, we comprehensively evaluate 10 prominent multimodal large language models. Currently, all models struggle, and the best results (Gemini 1.5 Pro) are around 33% accuracy, highlighting the severe deficiency of these techniques in egocentric QA assistance. Our further investigations suggest that precise temporal grounding and multi-frame reasoning, along with high resolution and auxiliary scene-text inputs, are key for better performance. With thorough analyses and heuristic suggestions, we hope EgoTextVQA can serve as a solid testbed for research in egocentric scene-text QA assistance.

Touch contact and pressure are essential for understanding how humans interact with objects and offer insights that benefit applications in mixed reality and robotics. Estimating these interactions from an egocentric camera perspective is challenging, largely due to the lack of comprehensive datasets that provide both hand poses and pressure annotations. In this paper, we present EgoPressure, an egocentric dataset that is annotated with high-resolution pressure intensities at contact points and precise hand pose meshes, obtained via our multi-view, sequence-based optimization method. We introduce baseline models for estimating applied pressure on external surfaces from RGB images, both with and without hand pose information, as well as a joint model for predicting hand pose and the pressure distribution across the hand mesh.Our experiments show that pressure and hand pose complement each other in understanding hand-object interactions.

We introduce EgoLife, a project to develop an egocentric life assistant that accompanies and enhances personal efficiency through AI-powered wearable glasses. To lay the foundation for this assistant, we conducted a comprehensive data collection study where six participants lived together for one week, continuously recording their daily activities - including discussions, shopping, cooking, socializing, and entertainment - using AI glasses for multimodal egocentric video capture, along with synchronized third-person-view video references. This effort resulted in the EgoLife Dataset, a comprehensive 300-hour egocentric, interpersonal, multiview, and multimodal daily life dataset with intensive annotation. Leveraging this dataset, we introduce EgoLifeQA, a suite of long-context, life-oriented question-answering tasks designed to provide meaningful assistance in daily life by addressing practical questions such as recalling past relevant events, monitoring health habits, and offering personalized recommendations. To address the key technical challenges of (1) developing robust visual-audio models for egocentric data, (2) enabling identity recognition, and (3) facilitating long-context question answering over extensive temporal information, we introduce EgoButler, an integrated system comprising EgoGPT and EgoRAG. EgoGPT is an omni-modal model trained on egocentric datasets, achieving state-of-the-art performance on egocentric video understanding. EgoRAG is a retrieval-based component that supports answering ultra-long-context questions. Our experimental studies verify their working mechanisms and reveal critical factors and bottlenecks, guiding future improvements. By releasing our datasets, models, and benchmarks, we aim to stimulate further research in egocentric AI assistants.

As wearable devices become more prevalent, understanding the user's motion is crucial for improving contextual AI systems. We introduce EgoLM, a versatile framework designed for egocentric motion understanding using multi-modal data. EgoLM integrates the rich contextual information from egocentric videos and motion sensors afforded by wearable devices. It also combines dense supervision signals from motion and language, leveraging the vast knowledge encoded in pre-trained large language models (LLMs). EgoLM models the joint distribution of egocentric motions and natural language using LLMs, conditioned on observations from egocentric videos and motion sensors. It unifies a range of motion understanding tasks, including motion narration from video or motion data, as well as motion generation from text or sparse sensor data. Unique to wearable devices, it also enables a novel task to generate text descriptions from sparse sensors. Through extensive experiments, we validate the effectiveness of EgoLM in addressing the challenges of under-constrained egocentric motion learning, and demonstrate its capability as a generalist model through a variety of applications.

This work focuses on tracking and understanding human motion using consumer wearable devices, such as VR/AR headsets, smart glasses, cellphones, and smartwatches. These devices provide diverse, multi-modal sensor inputs, including egocentric images, and 1-3 sparse IMU sensors in varied combinations. Motion descriptions can also accompany these signals. The diverse input modalities and their intermittent availability pose challenges for consistent motion capture and understanding. In this work, we present Ego4o (o for omni), a new framework for simultaneous human motion capture and understanding from multi-modal egocentric inputs. This method maintains performance with partial inputs while achieving better results when multiple modalities are combined. First, the IMU sensor inputs, the optional egocentric image, and text description of human motion are encoded into the latent space of a motion VQ-VAE. Next, the latent vectors are sent to the VQ-VAE decoder and optimized to track human motion. When motion descriptions are unavailable, the latent vectors can be input into a multi-modal LLM to generate human motion descriptions, which can further enhance motion capture accuracy. Quantitative and qualitative evaluations demonstrate the effectiveness of our method in predicting accurate human motion and high-quality motion descriptions.

For the deployment of neural networks in resource-constrained environments, prior works have built lightweight architectures with convolution and attention for capturing local and global dependencies, respectively. Recently, the state space model (SSM) has emerged as an effective operation for global interaction with its favorable linear computational cost in the number of tokens. To harness the efficacy of SSM, we introduce Efficient Vision Mamba (EfficientViM), a novel architecture built on hidden state mixer-based state space duality (HSM-SSD) that efficiently captures global dependencies with further reduced computational cost. With the observation that the runtime of the SSD layer is driven by the linear projections on the input sequences, we redesign the original SSD layer to perform the channel mixing operation within compressed hidden states in the HSM-SSD layer. Additionally, we propose multi-stage hidden state fusion to reinforce the representation power of hidden states and provide the design to alleviate the bottleneck caused by the memory-bound operations. As a result, the EfficientViM family achieves a new state-of-the-art speed-accuracy trade-off on ImageNet-1k, offering up to a 0.7% performance improvement over the second-best model SHViT with faster speed. Further, we observe significant improvements in throughput and accuracy compared to prior works, when scaling images or employing distillation training. Code is available at https://github.com/mlvlab/EfficientViM.

While multimodal large language models demonstrate strong performance in complex reasoning tasks, they pose significant challenges related to model complexity during deployment, especially for resource-limited devices. In this paper, we propose an automatic pruning method for large vision-language models to enhance the efficiency of multimodal reasoning. Conventional methods rely on the training data of the original model to select the proper pruning ratio for different network components. However, these methods are impractical for large vision-language models due to the unaffordable search costs caused by web-scale training corpus. In contrast, our approach only leverages a small number of samples to search for the desired pruning policy by maximizing its generalization ability on unknown training data while maintaining the model accuracy, which enables the achievement of an optimal trade-off between accuracy and efficiency for large visual language models. Specifically, we formulate the generalization gap of the pruning strategy using the structural risk minimization principle. Based on both task performance and generalization capability, we iteratively search for the optimal pruning policy within a given search space and optimize the vision projector to evolve the search space with higher upper bound of performance. We conduct extensive experiments on the ScienceQA, Vizwiz, MM-vet, and LLaVA-Bench datasets for the task of visual question answering. Using only 64 samples for pruning policy search, EfficientLLaVA achieves an accuracy of 83.05% on ScienceQA, along with a x1.8 speedup compared to the dense LLaVA-v1.5-7B model.

Convolutional neural networks (CNNs) and Vision Transformers (ViTs) have achieved excellent performance in image restoration. While ViTs generally outperform CNNs by effectively capturing long-range dependencies and input-specific characteristics, their computational complexity increases quadratically with image resolution. This limitation hampers their practical application in high-resolution image restoration. In this paper, we propose a simple yet effective visual state space model (EVSSM) for image deblurring, leveraging the benefits of state space models (SSMs) to visual data. In contrast to existing methods that employ several fixed-direction scanning for feature extraction, which significantly increases the computational cost, we develop an efficient visual scan block that applies various geometric transformations before each SSM-based module, capturing useful non-local information and maintaining high efficiency. In addition, to more effectively capture and represent local information, we propose an efficient discriminative frequency domain-based feedforward network (EDFFN) which can effectively estimate useful frequency information for latent clear image restoration. Extensive experimental results show that the proposed EVSSM performs favorably against state-of-the-art methods on benchmark datasets and real-world images.

Latency is a key driver for real-time rendering applications, making super-resolution techniques increasingly popular to accelerate rendering processes. In contrast to existing methods that directly concatenate low-resolution frames and G-buffers as input without discrimination, we develop an asymmetric UNet-based super-resolution network with decoupled G-buffer guidance, dubbed RDG, to facilitate the spatial and temporal feature exploration for minimizing performance overheads and latency.We first propose a dynamic feature modulator (DFM) to selectively encode the spatial information to capture precise structural information.We then incorporate auxiliary G-buffer information to guide the decoder to generate detail-rich, temporally stable results.Specifically, we adopt a high-frequency feature booster (HFB) to adaptively transfer the high-frequency information from the normal and bidirectional reflectance distribution function (BRDF) components of the G-buffer, enhancing the details of the generated results.To further enhance the temporal stability, we design a cross-frame temporal refiner (CTR) with depth and motion vector constraints to aggregate the previous and current frames.Extensive experimental results reveal that our proposed method is capable of generating high-quality and temporally stable results in real-time rendering.The proposed RDG-s produces 1080P rendering results on a RTX 3090 GPU with a speed of 126 FPS.

As a very common type of video, face videos often appear in movies, talk shows, live broadcasts, and other scenes. Real-world online videos are often plagued by degradations such as blurring and quantization noise, due to the high compression ratio caused by high communication costs and limited transmission bandwidth. These degradations have a particularly serious impact on face videos because the human visual system is highly sensitive to facial details.Despite the significant advancement in video face enhancement, current methods still suffer from i) long processing time and ii) inconsistent spatial-temporal visual effects (e.g., flickering). This study proposes a novel and efficient blind video face enhancement method to overcome the above two challenges, restoring high-quality videos from their compressed low-quality versions with an effective de-flickering mechanism. In particular, the proposed method develops upon a 3D-VQGAN backbone associated with spatial-temporal codebooks recording high-quality portrait features and residual-based temporal information. We develop a two-stage learning framework for the model. In Stage I, we learn the model with a regularizer mitigating the codebook collapse problem.In Stage II, we learn two transformers to lookup code from the codebooks and further update the encoder of low-quality videos.Experiments conducted on the VFHQ-Test dataset demonstrate that our method surpasses the current state-of-the-art blind face video restoration and de-flickering methods on both efficiency and effectiveness.

Pre-trained vision-language models provide a robust foundation for efficient transfer learning across various downstream tasks. In the field of video action recognition, mainstream approaches often introduce additional modules to capture temporal information. Although the additional modules increase the capacity of model, enabling it to better capture video-specific inductive biases, existing methods typically introduce a substantial number of new parameters and are prone to catastrophic forgetting of previously acquired generalizable knowledge. In this paper, we propose a parameter-efficient Multi-modal Spatio-Temporal Adapter (MSTA) to enhance the alignment between textual and visual representations, achieving a balance between generalizable knowledge and task-specific adaptation. Furthermore, to mitigate over-fitting and enhance generalizability, we introduce a spatio-temporal description-guided consistency constraint. This constraint involves providing template inputs (e.g., "a video of \ cls\ ") to the trainable language branch and LLM-generated spatio-temporal descriptions to the pre-trained language branch, enforcing output consistency between the branches. This approach reduces overfitting to downstream tasks and enhances the distinguishability of the trainable branch within the spatio-temporal semantic space. We evaluate the effectiveness of our approach across four tasks: zero-shot transfer, few-shot learning, base-to-novel generalization, and fully-supervised learning. Compared to many state-of-the-art methods, our MSTA achieves outstanding performance across all evaluations, while using only 2-7% of the trainable parameters in the original model.

Diffusion models have shown remarkable performance in image synthesis, but they demand extensive computational and memory resources for training, fine-tuning and inference. Although advanced quantization techniques have successfully minimized memory usage for inference, training and fine-tuning these quantized models still require large memory possibly due to dequantization for accurate computation of gradients and/or backpropagation for gradient-based algorithms. However, memory-efficient fine-tuning is particularly desirable for applications such as personalization that often must be run on edge devices like mobile phones with private data. In this work, we address this challenge by quantizing a diffusion model with personalization via Textual Inversion and by leveraging a zeroth-order optimization on personalization tokens without dequantization so that it does not require gradient and activation storage for backpropagation that consumes considerable memory. Since a gradient estimation using zeroth-order optimization is quite noisy for a single or a few images in personalization, we propose to denoise the estimated gradient by projecting it onto a subspace that is constructed with the past history of the tokens, dubbed Subspace Gradient. In addition, we investigated the influence of text embedding in image generation, leading to our proposed time steps sampling, dubbed Partial Uniform Timestep Sampling for sampling with effective diffusion timesteps. Our method achieves comparable performance to prior methods in image and text alignment scores for personalizing Stable Diffusion with only forward passes while reducing training memory demand up to 8.2x. Project page: https://ignoww.github.io/ZOODiP_project/

Most current video MLLMs rely on uniform frame sampling and image-level encoders, resulting in inefficient data processing and limited motion awareness. To address these challenges, we introduce EMA, an Efficient Motion-Aware video MLLM that utilizes compressed video structures as inputs. We propose a motion-aware GOP (Group of Pictures) encoder that fuses spatial and motion information within a GOP unit in the compressed video stream, generating compact, informative visual tokens. By integrating fewer but denser RGB frames with more but sparser motion vectors in this native slow-fast input architecture, our approach reduces redundancy and enhances motion representation. Additionally, we introduce MotionBench, a benchmark for evaluating motion understanding across four motion types: linear, curved, rotational, and contact-based. Experimental results show that EMA achieves state-of-the-art performance on both MotionBench and popular video question answering benchmarks, while reducing inference costs. Moreover, EMA demonstrates strong scalability, as evidenced by its competitive performance on long video understanding benchmarks.

Efficient tokenization of videos remains a challenge in training vision models that can process long videos. One promising direction is to develop a tokenizer that can encode long video clips, as it would enable the tokenizer to leverage the temporal coherence of videos better for tokenization. However, training existing tokenizers on long videos often incurs a huge training cost as they are trained to reconstruct all the frames at once. In this paper, we introduce CoordTok, a video tokenizer that learns a mapping from coordinate-based representations to the corresponding patches of input videos, inspired by recent advances in 3D generative models. In particular, CoordTok encodes a video into factorized triplane representations and reconstructs patches that correspond to randomly sampled (x,y,t) coordinates. This allows for training large tokenizer models directly on long videos without requiring excessive training resources. Our experiments show that CoordTok can drastically reduce the number of tokens for encoding long video clips. For instance, CoordTok can encode a 128-frame video with 128x128 resolution into 1280 tokens, while baselines need 6144 or 8192 tokens to achieve similar reconstruction quality. We further show that this efficient video tokenization enables memory-efficient training of a diffusion transformer that can generate 128 frames at once.

Recent advances in diffusion generative models have yielded remarkable progress. While the quality of generated content continues to improve, these models have grown considerably in size and complexity. This increasing computational burden poses significant challenges, particularly in resource-constrained deployment scenarios such as mobile devices. The combination of model pruning and knowledge distillation has emerged as a promising solution to reduce computational demands while preserving generation quality. However, this technique inadvertently propagates undesirable behaviors, including the generation of copyrighted content and unsafe concepts, even when such instances are absent from the fine-tuning dataset. In this paper, we propose a novel bilevel optimization framework for pruned diffusion models that consolidates the fine-tuning and unlearning processes into a unified phase. Our approach maintains the principal advantages of distillation--namely, efficient convergence and style transfer capabilities--while selectively suppressing the generation of unwanted content. This plug-in framework is compatible with various pruning and concept unlearning methods, facilitating efficient, safe deployment of diffusion models in controlled environments.