CVPR 2025 Papers

Neural networks have revolutionized numerous fields with their exceptional performance, yet they remain susceptible to adversarial attacks through subtle perturbations. While diffusion-based purification methods like DiffPure offer promising defense mechanisms, their computational overhead presents a significant practical limitation. In this paper, we introduce One Step Control Purification (OSCP), a novel defense framework that achieves robust adversarial purification in a single Neural Function Evaluation (NFE) within diffusion models. We propose Gaussian Adversarial Noise Distillation (GAND) as the distillation objective and Controlled Adversarial Purification (CAP) as the inference pipeline, which makes OSCP demonstrate remarkable efficiency while maintaining defense efficacy. Our proposed GAND addresses a fundamental tension between consistency distillation and adversarial perturbation, bridging the gap between natural and adversarial manifolds in the latent space, while remaining computationally efficient through Parameter-Efficient Fine-Tuning (PEFT) methods such as LoRA, eliminating the high computational budget request from full parameter fine-tuning. The CAP guides the purification process through the unlearnable edge detection operator calculated by the input image as an extra prompt, effectively preventing the purified images from deviating from their original appearance when using large purification steps. Our experimental results on ImageNet showcase OSCP's superior performance, achieving a 74.19% defense success rate with merely 0.1s per purification --- a 100-fold speedup compared to conventional approaches.

3D scene understanding is vital for applications in autonomous driving, robotics, and augmented reality. However, scene understanding based on 3D Gaussian Splatting faces three key challenges: (i) an imbalance between appearance and semantics, (ii) inconsistencies in object boundaries, and (iii) difficulties with top-down instance segmentation. To address these challenges, we propose InstanceGaussian, a method that jointly learns appearance and semantic features while adaptively aggregating instances. Our contributions are as follows: (i) a new Semantic-Scaffold-GS representation to improve feature representation and boundary delineation, (ii) a progressive training strategy for enhanced stability and segmentation, and (iii) a category-agnostic, bottom-up instance aggregation approach for better segmentation. Experimental results demonstrate that our approach achieves state-of-the-art performance in category-agnostic, open-vocabulary 3D point-level segmentation, validating the effectiveness of our proposed method.

Text-to-video generation has evolved rapidly in recent years, delivering remarkable results. Training typically relies on video-caption paired data, which plays a crucial role in enhancing generation performance. However, current video captions often suffer from insufficient details, hallucinations and imprecise motion depiction, affecting the fidelity and consistency of generated videos. In this work, we propose a novel instance-aware structured caption framework, termed \mathtt InstanceCap , to achieve instance-level and fine-grained video caption for the first time. Based on this scheme, we design an auxiliary models cluster to convert original video into instances to enhance instance fidelity. Video instances are further used to refine dense prompts into structured phrases, achieving concise yet precise descriptions. Furthermore, a 22K \mathtt InstanceVid dataset is curated for training, and an enhancement pipeline that tailored to \mathtt InstanceCap structure is proposed for inference. Experimental results demonstrate that our proposed \mathtt InstanceCap significantly outperform previous models, ensuring high fidelity between captions and videos while reducing hallucinations.

Active learning (AL) is a label-efficient machine learning paradigm that focuses on selectively annotating high-value instances to maximize learning efficiency. Its effectiveness can be further enhanced by incorporating weak supervision, which uses rough yet cost-effective annotations instead of exact (i.e., full) but expensive annotations. We introduce a novel AL framework, Instance-wise Supervision-Level Optimization (ISO), which not only selects the instances to annotate but also determines their optimal annotation level within a fixed annotation budget. Its optimization criterion leverages the value-to-cost ratio (VCR) of each instance while ensuring diversity among the selected instances. In classification experiments, ISO consistently outperforms traditional AL methods and surpasses a state-of-the-art AL approach that combines full and weak supervision, achieving higher accuracy at a lower overall cost.

Despite encouraging progress in 3D scene understanding, it remains challenging to develop an effective Large Multi-modal Model (LMM) that is capable of understanding and reasoning in complex 3D environments. Most previous methods typically encode 3D point and 2D image features separately, neglecting interactions between 2D semantics and 3D object properties, as well as the spatial relationships within the 3D environment. This limitation not only hinders comprehensive representations of 3D scene, but also compromises training and inference efficiency. To address these challenges, we propose a unified Instance-aware 3D Large Multi-modal Model (Inst3D-LMM) to deal with multiple 3D scene understanding tasks simultaneously. To obtain the fine-grained instance-level visual tokens, we first introduce a novel Multi-view Cross-Modal Fusion (MCMF) module to inject the multi-view 2D semantics into their corresponding 3D geometric features. For scene-level relation-aware tokens, we further present a 3D Instance Spatial Relation (3D-ISR) module to capture the intricate pairwise spatial relationships among objects. Additionally, we perform end-to-end multi-task instruction tuning simultaneously without the subsequent task-specific fine-tuning. Extensive experiments demonstrate that our approach outperforms the state-of-the-art methods across 3D scene understanding, reasoning and grounding tasks. Source code is available at: https://github.com/hanxunyu/Inst3D-LMM.

Recent 3D Instance Segmentation methods typically encode hundreds of instance-wise candidates with instance-specific information in various ways and refine them into final masks. However, they have yet to fully explore the benefit of these candidates. They overlook the valuable cues encoded in multiple candidates that represent different parts of the same instance, resulting in fragments. Also, they often fail to capture the precise spatial range of 3D instances, primarily due to inherent noises from sparse and unordered point clouds. In this work, to address these challenges, we propose IKNE, a novel instance-wise knowledge enhancement approach. We first introduce an Instance-wise Knowledge Aggregation (IKA) to associate scattered single instance details by optimizing correlations among candidates representing the same instance. Moreover, we present an Instance-wise Structural Guidance (ISG) to enhance the spatial understanding of candidates using structural cues from ambiguity-reduced features. Here, we utilize a simple yet effective truncated singular value decomposition algorithm to minimize inherent noises of 3D features. In our extensive experiments on large-scale datasets, ScanNetV2, ScanNet200, S3DIS, and STPLS3D, IKNE outperforms existing works. We validate the effectiveness of our modules in both kernel-based and transformer-based architectures.

In this paper, we focus on the task of instruction-based image editing. Previous works like InstructPix2Pix, InstructDiffusion, and SmartEdit have explored end-to-end editing. However, two limitations still remain: First, existing datasets suffer from low resolution, poor background consistency, and overly simplistic instructions. Second, current approaches mainly condition on the text while the rich image information is underexplored, therefore inferior in complex instruction following and maintaining background consistency. Targeting these issues, we first curated the AdvancedEdit dataset using a novel data construction pipeline, formulating a large-scale dataset with high visual quality, complex instructions, and good background consistency. Then, to further inject the rich image information, we introduce a two-stream bridging mechanism utilizing both the textual and visual features reasoned by the powerful Multimodal Large Language Models (MLLM) to guide the image editing process more precisely. Extensive results demonstrate that our approach, InsightEdit, achieves state-of-the-art performance, excelling in complex instruction following and maintaining high background consistency with the original image.

Large Language Models (LLMs) demonstrate enhanced capabilities and reliability by reasoning more, evolving from Chain-of-Thought prompting to product-level solutions like OpenAI o1. Despite various efforts to improve LLM reasoning, high-quality long-chain reasoning data and optimized training pipelines still remain inadequately explored in vision-language tasks. In this paper, we present Insight-V, an early effort to 1) scalably produce long and robust reasoning data for complex multi-modal tasks, and 2) an effective training pipeline to enhance the reasoning capabilities of multi-modal large language models (MLLMs). Specifically, to create long and structured reasoning data without human labor, we design a two-step pipeline with a progressive strategy to generate sufficiently long and diverse reasoning paths and a multi-granularity assessment method to ensure data quality. We observe that directly supervising MLLMs with such long and complex reasoning data will not yield ideal reasoning ability. To tackle this problem, we design a multi-agent system consisting of a reasoning agent dedicated to performing long-chain reasoning and a summary agent trained to judge and summarize reasoning results. We further incorporate an iterative DPO algorithm to enhance the reasoning agent's generation stability and quality. Based on the popular LLaVA-NeXT model, our method shows an average improvement of 7.5% across seven challenging multi-modal benchmarks requiring visual reasoning. We also achieve a 4.2% improvement on a stronger base MLLM, highlighting the potential to further advance state-of-the-art models. Benefiting from our multi-agent system, Insight-V can also easily maintain or improve performance on perception-focused multi-modal tasks. We will make our data and code publicly available to promote future research in this emerging field.

Despite exhibiting impressive performance in synthesizing lifelike personalized 3D talking heads, prevailing methods based on radiance fields suffer from high demands for training data and time for each new identity. This paper introduces InsTaG, a 3D talking head synthesis framework that allows a fast learning of realistic personalized 3D talking head from few training data. Built upon a lightweight 3DGS person-specific synthesizer with universal motion priors, InsTaG achieves high-quality and fast adaptation while preserving high-level personalization and efficiency. As preparation, we first propose an Identity-Free Pre-training strategy that enables the pre-training of the person-specific model and encourages the collection of universal motion priors from long-video data corpus. To fully exploit the universal motion priors to learn an unseen new identity, we then present a Motion-Aligned Adaptation strategy to adaptively align the target head to the pre-trained field, and constrain a robust dynamic head structure under few training data. Experiments demonstrate our outstanding performance and efficiency under various data scenarios to render high-quality personalized talking heads. Project page: https://fictionarry.github.io/InsTaG/.

We present Infinity, a Bitwise Visual AutoRegressive Modeling capable of generating high-resolution, photorealistic images following language instruction. Infinity refactors visual autoregressive model under a bitwise token prediction framework with an infinite-vocabulary classifier and bitwise self-correction mechanism. By theoretically expanding the tokenizer vocabulary size to infinity in Transformer, our method significantly unleashes powerful scaling capabilities to infinity compared to vanilla VAR. Extensive experiments indicate Infinity outperforms AutoRegressive Text-to-Image models by large margins, matches or surpasses leading diffusion models. Without extra optimization, Infinity generates a 1024x1024 image in 0.8s, 2.6xfaster than SD3-Medium, making it the fastest Text-to-Image model. Models and codes will be released to promote the further exploration of Infinity for visual generation.

Federated Learning (FL), a privacy-preserving decentralized machine learning framework, has been shown to be vulnerable to backdoor attacks. Current research primarily focuses on the Single-Label Backdoor Attack (SBA), wherein adversaries share a consistent target. However, a critical fact is overlooked: adversaries may be non-cooperative, have distinct targets, and operate independently, which exhibits a more practical scenario called Multi-Label Backdoor Attack (MBA). Unfortunately, prior works are ineffective in the MBA scenario since non-cooperative attackers exclude each other. In this work, we conduct an in-depth investigation to uncover the inherent constraints of the exclusion: similar backdoor mappings are constructed for different targets, resulting in conflicts among backdoor functions. To address this limitation, we propose Mirage, the first non-cooperative MBA strategy in FL that allows attackers to inject effective and persistent backdoors into the global model without collusion by constructing in-distribution (ID) backdoor mapping. Specifically, we introduce an adversarial adaptation method to bridge the backdoor features and the target distribution in an ID manner. Additionally, we further leverage a constrained optimization method to ensure the ID mapping survives in the global training dynamics. Extensive evaluations demonstrate that Mirage outperforms various state-of-the-art attacks and bypasses existing defenses, achieving an average ASR greater than 97% and maintaining over 90% after 900 rounds. This work aims to alert researchers to this potential threat and inspire the design of effective defense mechanisms. Code has been made open-source.

Spiking Neural Networks (SNNs) have emerged as a promising substitute for Artificial Neural Networks (ANNs) due to their advantages of fast inference and low power consumption. However, the lack of efficient training algorithms has hindered their widespread adoption. Even efficient ANN-SNN conversion methods necessitate quantized training of ANNs to enhance the effectiveness of the conversion, incurring additional training costs. To address these challenges, we propose an efficient ANN-SNN conversion framework with only inference scale complexity. The conversion framework includes a local threshold balancing algorithm, which enables efficient calculation of the optimal thresholds and fine-grained adjustment of the threshold value by channel-wise scaling. We also introduce an effective delayed evaluation strategy to mitigate the influence of the spike propagation delays. We demonstrate the scalability of our framework in typical computer vision tasks: image classification, semantic segmentation, object detection, and video classification. Our algorithm outperforms existing methods, highlighting its practical applicability and efficiency. Moreover, we have evaluated the energy consumption of the converted SNNs, demonstrating their superior low-power advantage compared to conventional ANNs. This approach simplifies the deployment of SNNs by leveraging open-source pre-trained ANN models, enabling fast, low-power inference with negligible performance reduction.

3D Gaussian Splatting (3DGS) has shown impressive performance in scene reconstruction, offering high rendering quality and rapid rendering speed with short training time. However, it often yields unsatisfactory results when applied to indoor scenes due to its poor ability to learn geometries without enough textural information. In this paper, we propose a new 3DGS-based method "IndoorGS", that leverages the commonly found yet important geometric cues in indoor scenes to improve the reconstruction quality. Specifically, we first extract 2D lines from input images and fuse them into 3D line cues via feature-based matching, which can provide a structural understanding of the target scene. We then apply the statistical outlier removal to refine Structure-from-Motion (SfM) points, ensuring robust cues in texture-rich areas. Based on these two types of cues, we further extract reliable 3D plane cues for textureless regions. Such geometric information will be utilized not only for initialization but also in the realization of a geometric-cue-guided adaptive density control (ADC) strategy. The proposed ADC approach is grounded in the principle of divide-and-conquer and optimizes the use of each type of geometric cues to enhance overall reconstruction performance. Extensive experiments on multiple indoor datasets show that our method can deliver much more accurate geometry and higher rendering quality for indoor scenes than existing 3DGS approaches.

Existing progress in object keypoint estimation primarily benefits from the conventional supervised learning paradigm based on numerous data labeled with pre-defined keypoints. However, these well-trained models can hardly detect the undefined new keypoints in test time, which largely hinders their feasibility for diverse downstream tasks. To handle this, various solutions are explored but still suffer from either limited generalizability or transferability. Therefore, in this paper, we explore a novel keypoint learning paradigm in that we only annotate new keypoints in the new data and incrementally train the model, without retaining any old data, called Incremental object Keypoint Learning (IKL). A two-stage learning scheme as a novel baseline tailored to IKL is developed. In the first Knowledge Association stage, given the data labeled with only new keypoints, an auxiliary KA-Net is trained to automatically associate the old keypoints to these new ones based on their spatial and intrinsic anatomical relations. In the second Mutual Promotion stage, based on a keypoint-oriented spatial distillation loss, we jointly leverage the auxiliary KA-Net and the old model for knowledge consolidation to mutually promote the estimation of all old and new keypoints. Owing to the investigation of the correlations between new and old keypoints, our proposed method can not just effectively mitigate the catastrophic forgetting of old keypoints, but may even further improve the estimation of the old ones and achieve a positive transfer beyond anti-forgetting. Such an observation has been solidly verified by extensive experiments on different keypoint datasets, where our method exhibits superiority in alleviating the forgetting issue and boosting performance while enjoying labeling efficiency even under the low-shot data regime.

Leveraging Large Language Models (LLMs) for text representation has achieved significant success, but the exploration of using Multimodal LLMs (MLLMs) for multimodal representation remains limited. Previous MLLM-based representation studies have primarily focused on unifying the embedding space while neglecting the importance of multimodal alignment. As a result, their cross-modal retrieval performance falls markedly behind that of the CLIP series models. To address this, in our work, we 1) construct DeKon5M, a contrastive learning dataset enriched with dense multimodal knowledge, which efficiently enhances multimodal alignment capabilities in representation tasks. 2) design a framework for training unified representation on MLLMs. Building upon this unified representation framework and the dense knowledge dataset DeKon5M, we developed the dense knowledge representation model DeKR on Qwen2VL. Through extensive quantitative and qualitative experiments, our results demonstrate that DeKR not only aligns text, image, video, and text-image combinations within a unified embedding space but also achieves cross-modal retrieval performance comparable to SoTA CLIP series models. This fully validates the effectiveness of our approach and provides new insights for multimodal representation research.

Brain tumor segmentation plays a crucial role in clinical diagnosis, yet the frequent unavailability of certain MRI modalities poses a significant challenge. In this paper, we introduce the Learnable Sorting State Space Model (LS3M), a novel framework designed to maximize the utilization of available modalities for brain tumor segmentation. LS3M excels at efficiently modeling long-range dependencies based on the Mamba design, while incorporating differentiable permutation matrices that reorder input sequences based on modality-specific characteristics. This dynamic reordering ensures that critical spatial inductive biases and long-range semantic correlations inherent in 3D brain MRI are preserved, which is crucial for imcomplete multi-modal brain tumor segmentation.Once the input sequences are reordered using the generated permutation matrix, the Series State Space Model (S3M) block models the relationships between them, capturing both local and long-range dependencies. This enables effective representation of intra-modal and inter-modal relationships, significantly improving segmentation accuracy. Extensive experiments on the BraTS2018 and BraTS2020 datasets demonstrate that LS3M outperforms existing methods, offering a robust solution for brain tumor segmentation, particularly in scenarios with missing modalities.

In incomplete multi-view multi-label learning scenarios, it is crucial to use the incomplete multi-view data to extract consistent and specific representations from different data sources and to fully exploit the missing label information. However, most previous approaches ignore the separation problem between view-shared and specific information. To address this problem, in this paper, we propose a method that can separate view-consistent features from view-specific features under the Variational Autoencoder (VAE) framework. Specifically, we first introduce cross-view reconstruction to capture view-consistent features and extract shared information from different views through unsupervised pre-training. Subsequently, we develop a disentangling module to learn specific features by minimizing the variational upper bound of mutual information between consistent and specific features. Finally, we utilize prior label relevance information derived from training data to guide the learning of the distribution of label semantic embeddings, aggregating relevant semantic embeddings and maintaining the label relevance topology in the semantic space. In extensive experiments, our model outperforms existing state-of-the-art algorithms on several real-world datasets, which fully validates its strong adaptability to missing views and labels.

Implicit neural representation and explicit 3D Gaussian Splatting (3D-GS) for novel view synthesis have achieved remarkable progress with frame-based camera (e.g. RGB and RGB-D cameras) recently. Compared to frame-based camera, a novel type of bio-inspired visual sensor, i.e. event camera, has demonstrated advantages in high temporal resolution, high dynamic range, low power consumption and low latency, which make it being favored for many robotic applications. In this work, we present IncEventGS, an incremental 3D Gaussian Splatting reconstruction algorithm with a single event camera, without the assumption of known camera poses. To recover the 3D scene representation incrementally, we exploit the tracking and mapping paradigm of conventional SLAM pipelines for IncEventGS. Given the incoming event stream, the tracker first estimates an initial camera motion based on prior reconstructed 3D-GS scene representation. The mapper then jointly refines both the 3D scene representation and camera motion based on the previously estimated motion trajectory from the tracker. The experimental results demonstrate that IncEventGS delivers superior performance compared to prior NeRF-based methods and other related baselines, even we do not have the ground-truth camera poses. Furthermore, our method can also deliver better performance compared to state-of-the-art event visual odometry methods in terms of camera motion estimation.

Without using explicit reward, direct preference optimization (DPO) employs paired human preference data to fine-tune generative models, a method that has garnered considerable attention in large language models (LLMs). However, exploration of aligning text-to-image (T2I) diffusion models with human preferences remains limited. In comparison to supervised fine-tuning, existing methods that align diffusion model suffer from low training efficiency and subpar generation quality due to the long Markov chain process and the intractability of the reverse process. To address these limitations, we introduce DDIM-InPO, an efficient method for direct preference alignment of diffusion models. Our approach conceptualizes diffusion model as a single-step generative model, allowing us to fine-tune the outputs of specific latent variables selectively. In order to accomplish this objective, we first assign implicit rewards to any latent variable directly via a reparameterization technique. Then we construct an Inversion technique to estimate appropriate latent variables for preference optimization. This modification process enables the diffusion model to only fine-tune the outputs of latent variables that have a strong correlation with the preference dataset. Experimental results indicate that our DDIM-InPO achieves state-of-the-art performance with just 400 steps of fine-tuning, surpassing all preference aligning baselines for T2I diffusion models in human preference evaluation tasks.