CVPR 2025 Papers

Recent advances in generalizable 3D Gaussian Splatting have demonstrated promising results in real-time high-fidelity rendering without per-scene optimization, yet existing approaches still struggle to handle unfamiliar visual content during inference on novel scenes due to limited generalizability. To address this challenge, we introduce MonoSplat, a novel framework that leverages rich visual priors from pre-trained monocular depth foundation models for robust Gaussian reconstruction. Our approach consists of two key components: a Mono-Multi Feature Adapter that transforms monocular features into multi-view representations, coupled with an Integrated Gaussian Prediction module that effectively fuses both feature types for precise Gaussian generation. Through the Adapter's lightweight attention mechanism, features are seamlessly aligned and aggregated across views while preserving valuable monocular priors, enabling the Prediction module to generate Gaussian primitives with accurate geometry and appearance. Through extensive experiments on diverse real-world datasets, we convincingly demonstrate that MonoSplat achieves superior reconstruction quality and generalization capability compared to existing methods while maintaining computational efficiency with minimal trainable parameters. Codes are available at \href https://github.com/CUHK-AIM-Group/MonoSplat https://github.com/CUHK-AIM-Group/MonoSplat .

Current monocular 3D detectors are held back by the limited diversity and scale of real-world datasets. While data augmentation certainly helps, it's particularly difficult to generate realistic scene-aware augmented data for outdoor settings. Most current approaches to synthetic data generation focus on realistic object appearance through improved rendering techniques. However, we show that where and how objects are positioned is just as crucial for training effective 3D monocular detectors. The key obstacle lies in automatically determining realistic object placement parameters - including position, dimensions, and directional alignment when introducing synthetic objects into actual scenes. To address this, we introduce MonoPlace3D, a novel system that considers the 3D scene content to create realistic augmentations. Specifically, given a background scene, MonoPlace3D learns a distribution over plausible 3D bounding boxes. Subsequently, we render realistic objects and place them according to the locations sampled from the learned distribution. Our comprehensive evaluation on two standard datasets KITTI and NuScenes, demonstrates that MonoPlace3D significantly improves the accuracy of multiple existing monocular 3D detectors while being highly data efficient.

Monocular depth priors have been widely adopted by neural rendering in multi-view based tasks such as 3D reconstruction and novel view synthesis. However, due to the inconsistent prediction on each view, how to more effectively leverage monocular cues in a multi-view context remains a challenge. Current methods treat the entire estimated depth map indiscriminately, and use it as ground truth supervision, while ignoring the inherent inaccuracy and cross-view inconsistency in monocular priors. To resolve these issues, we propose MonoInstance, a general approach that explores the uncertainty of monocular depths to provide enhanced geometric priors for neural rendering and reconstruction. Our key insight lies in aligning each segmented instance depths from multiple views within a common 3D space, thereby casting the uncertainty estimation of monocular depths into a density measure within noisy point clouds. For high-uncertainty areas where depth priors are unreliable, we further introduce a constraint term that encourages the projected instances to align with corresponding instance masks on nearby views. MonoInstance is a versatile strategy which can be seamlessly integrated into various multi-view neural rendering frameworks. Our experimental results demonstrate that MonoInstance significantly improves the performance in both reconstruction and novel view synthesis under various benchmarks.

Perspective projection has been extensively utilized in monocular 3D object detection methods. It introduces geometric priors from 2D bounding boxes and 3D object dimensions to reduce the uncertainty of depth estimation. However, due to errors originating from the object's visual surface, the bounding box height often fails to represent the actual central height, which undermines the effectiveness of geometric depth. Direct prediction for the projected height unavoidably results in a loss of 2D priors, while multi-depth prediction with complex branches does not fully leverage geometric depth. This paper presents a Transformer-based monocular 3D object detection method called MonoDGP, which adopts perspective-invariant geometry errors to modify the projection formula. We also try to systematically discuss and explain the mechanisms and efficacy behind geometry errors, which serve as a simple but effective alternative to multi-depth prediction. Additionally, MonoDGP decouples the depth-guided decoder and constructs a 2D decoder only dependent on visual features, providing 2D priors and initializing object queries without the disturbance of 3D detection. To further optimize and fine-tune input tokens of the transformer decoder, we also introduce a Region Segmentation Head (RSH) that generates enhanced features and segment embeddings. Our monocular method demonstrates state-of-the-art performance on the KITTI benchmark without extra data. Code is available at https://github.com/PuFanqi23/MonoDGP.

Visual-Language Tracking (VLT) is emerging as a promising paradigm to bridge the human-machine performance gap. For single objects, VLT broadens the problem scope to text-driven video comprehension. Yet, this direction is still confined to 2D spatial extents, currently lacking the ability to deal with 3D tracking in the confines of monocular video. Unfortunately, advances in 3D tracking mainly rely on expensive sensor inputs, e.g., point clouds, depth measurements, radar. Absence of language counterpart for the outputs of these mildly democratized sensors in the literature also hinders VLT expansion to 3D tracking. Addressing that, we make the first attempt towards extending VLT to 3D tracking based on monocular video. We present a comprehensive framework, introducing (i) the Monocular-Video-based 3D Visual Language Tracking (Mono3DVLT) task, (ii) a large-scale dataset for the task, called Mono3DVLT-V2X, and (iii) a customized neural model for the task. Our dataset is carefully curated, leveraging a Large Langauge Model (LLM) followed by human verification, composing natural language descriptions for 79,158 video sequences aiming at single object tracking, providing 2D and 3D bounding box annotations. Our neural model, termed Mono3DVLT-MT, is the first targeted approach for the Mono3DVLT task. Comprising the pipeline of multi-modal feature extractor, visual-language encoder, tracking decoder and a tracking head, our model sets a strong baseline for the task on Mono3DVLT-V2X. Experimental results show that our method significantly outperforms existing techniques on the Mono3DVLT-V2X dataset. Our dataset and code are available in https://github.com/hongkai-wei/Mono3DVLT.

With the rapid proliferation of 3D devices and the shortage of 3D content, stereo conversion is attracting increasing attention. Recent works introduce pretrained Diffusion Models (DMs) into this task. However, due to the scarcity of large-scale training data and comprehensive benchmarks, the optimal methodologies for employing DMs in stereo conversion and the accurate evaluation of stereo effects remain largely unexplored. In this work, we introduce the Mono2Stereo dataset, providing high-quality training data and benchmark to support in-depth exploration of stereo conversion. With this dataset, we conduct an empirical study that yields two primary findings. 1) The differences between the left and right views are subtle, yet existing metrics consider overall pixels, failing to concentrate on regions critical to stereo effects. 2) Mainstream methods adopt either one-stage left-to-right generation or warp-and-inpaint pipeline, facing challenges of degraded stereo effect and image distortion respectively. Based on these findings, we introduce a new evaluation metric, Stereo Intersection-over-Union, which prioritizes disparity and achieves a high correlation with human judgments on stereo effect. Moreover, we propose a strong baseline model, harmonizing the stereo effect and image quality simultaneously, and notably surpassing current mainstream methods. Our code and data will be open-sourced to promote further research in stereo conversion.

In this paper, we focus on monolithic Multimodal Large Language Models (MLLMs) that integrate visual encoding and language decoding into a single LLM. In particular, we identify that existing pre-training strategies for monolithic MLLMs often suffer from unstable optimization or catastrophic forgetting. To address this issue, our core idea is to embed a new visual parameter space into a pre-trained LLM, thereby stably learning visual knowledge from noisy data while freezing the LLM. Based on this principle, we present Mono-InternVL, a novel monolithic MLLM that seamlessly integrates a set of visual experts via a multimodal mixture-of-experts structure. Moreover, we propose an innovative pre-training strategy to maximize the visual capability of Mono-InternVL, namely Endogenous Visual Pre-training (EViP). In particular, EViP is designed as a progressive learning process for visual experts, which aims to fully exploit the visual knowledge from noisy data to high-quality data. To validate our approach, we conduct extensive experiments on 16 benchmarks. Experimental results confirm the superior performance of Mono-InternVL over existing monolithic MLLMs on 13 of 16 multimodal benchmarks, e.g., +80 points over Emu3 on OCRBench. Compared to the modular baseline, i.e., InternVL-1.5, Mono-InternVL still retains comparable multimodal performance while reducing up to 67% first token latency. Our project is available at https://internvl.github.io/blog/2024-10-10-Mono-InternVL/.

Stereo matching recovers depth from image correspondences. Existing methods struggle to handle ill-posed regions with limited matching cues, such as occlusions and textureless areas. To address this, we propose MonSter, a novel method that leverages the complementary strengths of monocular depth estimation and stereo matching. MonSter integrates monocular depth and stereo matching into a dual-branch architecture to iteratively improve each other. Confidence-based guidance adaptively selects reliable stereo cues for monodepth scale-shift recovery, and utilizes explicit monocular depth priors to enhance stereo matching at ill-posed regions. Such iterative mutual enhancement enables MonSter to evolve monodepth priors from coarse object-level structures to pixel-level geometry, fully unlocking the potential of stereo matching. As shown in Fig.2, MonSter ranks 1st across five most commonly used leaderboards --- SceneFlow, KITTI 2012, KITTI 2015, Middlebury, and ETH3D. Achieving up to 49.5% improvements over the previous best method (Bad 1.0 on ETH3D). Comprehensive analysis verifies the effectiveness of MonSter in ill-posed regions. In terms of zero-shot generalization, MonSter significantly and consistently outperforms state-of-the-art methods across the board. Code will be released upon acceptance.

Today's most advanced vision-language models (VLMs) remain proprietary. The strongest open-weight models rely heavily on synthetic data from proprietary VLMs to achieve good performance, effectively distilling these closed VLMs into open ones. As a result, the community has been missing foundational knowledge about how to build performant VLMs from scratch. We present Molmo, a new family of VLMs that are state-of-the-art in their class of openness. Our key contribution is a collection of new datasets called PixMo, including a dataset of highly detailed image captions for pre-training, a free-form image Q&A dataset for fine-tuning, and an innovative 2D pointing dataset, all collected without the use of external VLMs. The success of our approach relies on careful modeling choices, a well-tuned training pipeline, and, most critically, the quality of our newly collected datasets. Our best-in-class 72B model not only outperforms others in the class of open weight and data models, but also outperforms larger proprietary models including Claude 3.5 Sonnet, and Gemini 1.5 Pro and Flash, second only to GPT-4o based on both academic benchmarks and on a large human evaluation. Our model weights, new datasets, and source code are available at https://molmo.allenai.org/blog.

Text-to-image person re-identification (ReID) aims to retrieve the images of an interested person based on textual descriptions. One main challenge for this task is the high cost in manually annotating large-scale databases, which affects the generalization ability of ReID models. Recent works handle this problem by leveraging Multi-modal Large Language Models (MLLMs) to describe pedestrian images automatically. However, the captions produced by MLLMs lack diversity in description styles. To address this issue, we propose a Human Annotator Modeling (HAM) approach to enable MLLMs to mimic the description styles of thousands of human annotators. Specifically, we first extract style features from human textual descriptions and perform clustering on them. This allows us to group textual descriptions with similar styles into the same cluster. Then, we employ a prompt to represent each of these clusters and apply prompt learning to mimic the description styles of different human annotators. Furthermore, we define a style feature space and perform uniform sampling in this space to obtain more diverse clustering prototypes, which further enriches the diversity of the MLLM-generated captions. Finally, we adopt HAM to automatically annotate a massive-scale database for text-to-image ReID. Extensive experiments on this database demonstrate that it significantly improves the generalization ability of ReID models.

Error detection in procedural activities is essential for consistent and correct outcomes in AR-assisted and robotic systems. Existing methods often focus on temporal ordering errors or rely on static prototypes to represent normal actions. However, these approaches typically overlook the common scenario where multiple, distinct actions are valid following a given sequence of executed actions. This leads to two issues: (1) the model cannot effectively detect errors using static prototypes when the inference environment or action execution distribution differs from training; and (2) the model may also use the wrong prototypes to detect errors if the ongoing action label is not the same as the predicted one. To address this problem, we propose an Adaptive Multiple Normal Action Representation (AMNAR) framework. AMNAR predicts all valid next actions and reconstructs their corresponding normal action representations, which are compared against the ongoing action to detect errors. Extensive experiments demonstrate that AMNAR achieves state-of-the-art performance, highlighting the effectiveness of AMNAR and the importance of modeling multiple valid next actions in error detection. The code is available at https://github.com/iSEE-Laboratory/AMNAR.

Model poisoning attacks are critical security threats to Federated Learning (FL). Existing model poisoning attacks suffer from two key limitations: 1) they achieve suboptimal effectiveness when defenses are deployed, and/or 2) they require knowledge of the model updates or local training data on genuine clients. In this work, we make a key observation that their suboptimal effectiveness arises from only leveraging model-update consistency among malicious clients within individual training rounds, making the attack effect self-cancel across training rounds. In light of this observation, we propose PoisonedFL, which enforces multi-round consistency among the malicious clients' model updates while not requiring any knowledge about the genuine clients. Our empirical evaluation on five benchmark datasets shows that PoisonedFL breaks eight state-of-the-art defenses and outperforms seven existing model poisoning attacks. Our study shows that FL systems are considerably less robust than previously thought, underlining the urgency for the development of new defense mechanisms. Our source code is available at https://github.com/xyq7/PoisonedFL/.

How can we troubleshoot a deep visual model, i.e., understand why it makes certain mistakes and further take action to correct its behavior? We design a Model Diagnosis and Correction system (MDC), an automated framework that analyzes the pattern of errors, proposes candidate causes of attributes, conducts hypothesis testing via attribute editing, and ultimately generates counterfactual training samples to improve the performance of the model. Unlike previous methods, in addition to the linguistic attributes, our method also incorporates the analysis for implicit causal attributes, those cannot to be accurately described by language. To achieve this, we propose an image editing module capable of leveraging both implicit and linguistic attributes to generate counterfactual images depicting error patterns and further experimentally validate causality relationships. Lastly, we enrich the training set with synthetic samples depicting verified causal attributes and retrain the model, further boosting accuracy and robustness. Extensive experiments on fine-grained classification and face security applications demonstrate the superiority of our approach in model diagnosis and correction. Specifically, we achieve an average relative improvement of 62.01% in HTER for face security application over state-of-the-art methods.

Anticipating the multimodality of future events lays the foundation for safe autonomous driving. However, multimodal motion prediction for traffic agents has been clouded by the lack of multimodal ground truth. Existing works predominantly adopt the winner-take-all training strategy to tackle this challenge, yet still suffer from limited trajectory diversity and uncalibrated mode confidence. While some approaches address these limitations by generating excessive trajectory candidates, they necessitate a post-processing stage to identify the most representative modes, a process lacking universal principles and compromising trajectory accuracy. We are thus motivated to introduce ModeSeq, a new multimodal prediction paradigm that models modes as sequences. Unlike the common practice of decoding multiple plausible trajectories in one shot, ModeSeq requires motion decoders to infer the next mode step by step, thereby more explicitly capturing the correlation between modes and significantly enhancing the ability to reason about multimodality. Leveraging the inductive bias of sequential mode prediction, we also propose the Early-Match-Take-All (EMTA) training strategy to diversify the trajectories further. Without relying on dense mode prediction or heuristic post-processing, ModeSeq considerably improves the diversity of multimodal output while attaining satisfactory trajectory accuracy, resulting in balanced performance on motion prediction benchmarks. Moreover, ModeSeq naturally emerges with the capability of mode extrapolation, which supports forecasting more behavior modes when the future is highly uncertain.

Existing neural head avatars methods have achieved significant progress in the image quality and motion range of portrait animation. However, these methods prioritize effectiveness over computational overhead. This paper presents MobilePortrait, a lightweight one-shot neural head avatars method that reduces learning complexity by integrating external knowledge into both the motion modeling and image synthesis, enabling real-time inference on mobile devices. Specifically, We introduce a mixed keypoint representation of explicit and implicit keypoints for precise motion modeling and equip it with precomputed visual features to enhance both facial and background synthesis. With these designs, our model can achieve state-of-the-art performance using UNet with low FLOPs as the backbone, requiring less than 1/10 the computational demand. It has been validated to reach speeds of over 50 FPS on mobile devices and support both video and audio-driven inputs.

Previous research on lightweight models has primarily focused on CNNs and Transformer-based designs. CNNs, with their local receptive fields, struggle to capture long-range dependencies, while Transformers, despite their global modeling capabilities, are limited by quadratic computational complexity in high-resolution scenarios. Recently, state-space models have gained popularity in the visual domain due to their linear computational complexity. Despite their low FLOPs, current lightweight Mamba-based models exhibit suboptimal throughput. In this work, we propose the MobileMamba framework, which balances efficiency and performance. We design a three-stage network to enhance inference speed significantly. At a fine-grained level, we introduce the Multi-Receptive Field Feature Interaction (MRFFI) module, comprising the Long-Range Wavelet Transform-Enhanced Mamba (WTE-Mamba), Efficient Multi-Kernel Depthwise Deconvolution (MK-DeConv), and Eliminate Redundant Identity components. This module integrates multi-receptive field information and enhances high-frequency detail extraction. Additionally, we employ training and testing strategies to further improve performance and efficiency. MobileMamba achieves up to 83.6% on Top-1, surpassing existing state-of-the-art methods which is maximum x21| faster than LocalVim on GPU. Extensive experiments on high-resolution downstream tasks demonstrate that MobileMamba surpasses current efficient models, achieving an optimal balance between speed and accuracy.

This paper introduces MobileH2R, a framework for learning generalizable vision-based human-to-mobile-robot (H2MR) handover skills. Unlike traditional fixed-base handovers, this task requires a mobile robot to reliably receive objects in a large workspace enabled by its mobility. Our key insight is that generalizable handover skills can be developed in simulators using high-quality synthetic data, without the need for real-world demonstrations. To achieve this, we propose a scalable pipeline for generating diverse synthetic full-body human motion data, an automated method for creating safe and imitation-friendly demonstrations, and an efficient 4D imitation learning method for distilling large-scale demonstrations into closed-loop policies with base-arm coordination. Experimental evaluations in both simulators and the real world show significant improvements (at least +15% success rate) over baseline methods in all cases. Experiments also validate that large-scale and diverse synthetic data greatly enhances robot learning, highlighting our scalable framework.

Visual encoders are fundamental components in vision-language models (VLMs), each showcasing unique strengths derived from various pre-trained visual foundation models. To leverage the various capabilities of these encoders, recent studies incorporate multiple encoders within a single VLM, leading to a considerable increase in computational cost. In this paper, we present Mixture-of-Visual-Encoder Knowledge Distillation (MoVE-KD), a novel framework that distills the unique proficiencies of multiple vision encoders into a single, efficient encoder model. Specifically, to mitigate conflicts and retain the unique characteristics of each teacher encoder, we employ low-rank adaptation (LoRA) and mixture-of-experts (MoEs) to selectively activate specialized knowledge based on input features, enhancing both adaptability and efficiency. To regularize the KD process and enhance performance, we propose an attention-based distillation strategy that adaptively weighs the different encoders and emphasizes valuable visual tokens, reducing the burden of replicating comprehensive but distinct features from multiple teachers. Comprehensive experiments on popular VLMs, such as LLaVA and LLaVA-NeXT, validate the effectiveness of our method. Our code is available at:https://github.com/hey-cjj/MoVE-KD.

We introduce 4D Motion Scaffolds (MoSca), a modern 4D reconstruction system designed to reconstruct and synthesize novel views of dynamic scenes from monocular videos captured casually in the wild. To address such a challenging and ill-posed inverse problem, we leverage prior knowledge from foundational vision models and lift the video data to a novel Motion Scaffold (MoSca) representation, which compactly and smoothly encodes the underlying motions/deformations. The scene geometry and appearance are then disentangled from the deformation field and are encoded by globally fusing the Gaussians anchored onto the MoSca and optimized via Gaussian Splatting. Additionally, camera focal length and poses can be solved using bundle adjustment without the need of any other pose estimation tools. Experiments demonstrate state-of-the-art performance on dynamic rendering benchmarks and its effectiveness on real videos.