ML + Vision Top-6 Agent Survey - CVPR 2025 - Page 2 of 5¶
- Venue: Computer Vision and Pattern Recognition
- Year: 2025
- Page: 2 / 5
- Papers: 31-60 / 124
Papers
4D LangSplat: 4D Language Gaussian Splatting via Multimodal Large Language Models Paper
Abstract
Learning 4D language fields to enable time-sensitive, open-ended language queries in dynamic scenes is essential for many real-world applications. While LangSplat successfully grounds CLIP features into 3D Gaussian representations, achieving precision and efficiency in 3D static scenes, it lacks the ability to handle dynamic 4D fields as CLIP, designed for static image-text tasks, cannot capture temporal dynamics in videos. Real-world environments are inherently dynamic, with object semantics evolving over time. Building a precise 4D language field necessitates obtaining pixel-aligned, object-wise video features, which current vision models struggle to achieve. To address these challenges, we propose 4D LangSplat, which learns 4D language fields to handle time-agnostic or time-sensitive open-vocabulary queries in dynamic scenes efficiently. 4D LangSplat bypasses learning the language field from vision features and instead learns directly from text generated from object-wise video captions via Multimodal Large Language Models (MLLMs). Specifically, we propose a multimodal object-wise video prompting method, consisting of visual and text prompts that guide MLLMs to generate detailed, temporally consistent, high-quality captions for objects throughout a video. These captions are encoded using a Large Language Model into high-quality sentence embeddings, which then serve as pixel-aligned, object-specific feature supervision, facilitating open-vocabulary text queries through shared embedding spaces. Recognizing that objects in 4D scenes exhibit smooth transitions across states, we further propose a status deformable network to model these continuous changes over time effectively. Our results across multiple benchmarks demonstrate that 4D LangSplat attains precise and efficient results for both timesensitive and time-agnostic open-vocabulary queries.
Claim
Learning 4D language fields to enable time-sensitive, open-ended language queries in dynamic scenes is essential for many real-world applications.
Global-Local Tree Search in VLMs for 3D Indoor Scene Generation Paper
Abstract
Large Vision-Language Models (VLMs), such as GPT-4, have achieved remarkable success across various fields. However, there are few studies on 3D indoor scene generation with VLMs. This paper considers this task as a planning problem subject to spatial and layout common sense constraints. To solve the problem with a VLM, we propose a new global-local tree search algorithm. Globally, the method places each object sequentially and explores multiple placements during each placement process, where the problem space is represented as a tree. To reduce the depth of the tree, we decompose the scene structure hierarchically, i.e. room level, region level, floor object level, and supported object level. The algorithm independently generates the floor objects in different regions and supported objects placed on different floor objects. Locally, we also decompose the sub-task, the placement of each object, into multiple steps. The algorithm searches the tree of problem space. To leverage the VLM model to produce positions of objects, we discretize the top-down view space as a dense grid and fill each cell with diverse emojis to make to cells distinct. We prompt the VLM with the emoji grid and the VLM produces a reasonable location for the object by describing the position with the name of emojis. The quantitative and qualitative experimental results illustrate our approach generates more plausible 3D scenes than state-of-the-art approaches. Our source code is available at https://github.com/dw-dengwei/TreeSearchGen.
Claim
Large Vision-Language Models (VLMs), such as GPT-4, have achieved remarkable success across various fields.
ECBench: Can Multi-modal Foundation Models Understand the Egocentric World? A Holistic Embodied Cognition Benchmark Paper
Abstract
The enhancement of generalization in robots by large vision-language models (LVLMs) is increasingly evident. Therefore, the embodied cognitive abilities of LVLMs based on egocentric videos are of great interest. However, current datasets for embodied video question answering lack comprehensive and systematic evaluation frameworks. Critical embodied cognitive issues, such as robotic self-cognition, dynamic scene perception, and hallucination, are rarely addressed. To tackle these challenges, we propose ECBench, a high-quality benchmark designed to systematically evaluate the embodied cognitive abilities of LVLMs. ECBench features a diverse range of scene video sources, open and varied question formats, and 30 dimensions of embodied cognition. To ensure quality, balance, and high visual dependence, ECBench uses class-independent meticulous human annotation and multi-round question screening strategies. Additionally, we introduce ECEval, a comprehensive evaluation system that ensures the fairness and rationality of the indicators. Utilizing ECBench, we conduct extensive evaluations of proprietary, open-source, and task-specific LVLMs. ECBench is pivotal in advancing the embodied cognitive capabilities of LVLMs, laying a solid foundation for developing reliable core models for embodied agents. All data and code is available at https://github.com/RhDang/ECBench.
Claim
The enhancement of generalization in robots by large vision-language models (LVLMs) is increasingly evident.
FireEdit: Fine-grained Instruction-based Image Editing via Region-aware Vision Language Model Paper
Abstract
Currently, instruction-based image editing methods have made significant progress by leveraging the powerful cross-modal understanding capabilities of vision language models (VLMs). However, they still face challenges in three key areas: 1) complex scenarios; 2) semantic consistency; and 3) fine-grained editing. To address these issues, we propose FireEdit, an innovative Fine-grained Instruction-based image editing framework that exploits a REgion-aware VLM. FireEdit is designed to accurately comprehend user instructions and ensure effective control over the editing process. Specifically, we enhance the fine-grained visual perception capabilities of the VLM by introducing additional region tokens. Relying solely on the output of the LLM to guide the diffusion model may lead to suboptimal editing results. Therefore, we propose a Time-Aware Target Injection module and a Hybrid Visual Cross Attention module. The former dynamically adjusts the guidance strength at various denoising stages by integrating timestep embeddings with the text embeddings. The latter enhances visual details for image editing, thereby preserving semantic consistency between the edited result and the source image. By combining the VLM enhanced with fine-grained region tokens and the time-dependent diffusion model, FireEdit demonstrates significant advantages in comprehending editing instructions and maintaining high semantic consistency. Extensive experiments indicate that our approach surpasses the state-of-the-art instruction-based image editing methods.
Claim
Currently, instruction-based image editing methods have made significant progress by leveraging the powerful cross-modal understanding capabilities of vision language models (VLMs).
Mamba as a Bridge: Where Vision Foundation Models Meet Vision Language Models for Domain-Generalized Semantic Segmentation Paper
Abstract
Vision Foundation Models (VFMs) and Vision-Language Models (VLMs) have gained traction in Domain Generalized Semantic Segmentation (DGSS) due to their strong generalization capabilities1. However, existing DGSS methods often rely exclusively on either VFMs or VLMs, overlooking their complementary strengths. VFMs (e.g., DINOv2) excel at capturing fine-grained features, while VLMs (e.g., CLIP) provide robust text alignment but struggle with coarse granularity. Despite their complementary strengths, effectively integrating VFMs and VLMs with attention mechanisms is challenging, as the increased patch tokens complicate long-sequence modeling. To address this, we propose MFuser, a novel Mamba-based fusion framework that efficiently combines the strengths of VFMs and VLMs while maintaining linear scalability in sequence length. MFuser consists of two key components: MVFuser, which acts as a co-adapter to jointly fine-tune the two models by capturing both sequential and spatial dynamics; and MTEnhancer, a hybrid attention-Mamba module that refines text embeddings by incorporating image priors. Our approach achieves precise feature locality and strong text alignment without incurring significant computational overhead. Extensive experiments demonstrate that MFuser significantly outperforms state-of-the-art DGSS methods, achieving 68.20 mIoU on synthetic-to-real and 71.87 mIoU on real-to-real benchmarks. The code is available at https://github.com/devinxzhang/MFuser.
Claim
Vision Foundation Models (VFMs) and Vision-Language Models (VLMs) have gained traction in Domain Generalized Semantic Segmentation (DGSS) due to their strong generalization capabilities1.
PhysVLM: Enabling Visual Language Models to Understand Robotic Physical Reachability Paper
Abstract
Understanding the environment and a robot’s physical reachability is crucial for task execution. While state-of-the-art vision-language models (VLMs) excel in environmental perception, they often generate inaccurate or impractical responses in embodied visual reasoning tasks due to a lack of understanding of robotic physical reachability. To address this issue, we propose a unified representation of physical reachability across diverse robots, i.e., Space-Physical Reachability Map (S-P Map), and PhysVLM, a vision-language model that integrates this reachability information into visual reasoning. Specifically, the S-P Map abstracts a robot’s physical reachability into a generalized spatial representation, independent of specific robot configurations, allowing the model to focus on reachability features rather than robot-specific parameters. Subsequently, PhysVLM extends traditional VLM architectures by incorporating an additional feature encoder to process the S-P Map, enabling the model to reason about physical reachability without compromising its general vision-language capabilities. To train and evaluate PhysVLM, we constructed a large-scale multi-robot dataset, Phys100K, and a challenging benchmark, EQA-phys, which includes tasks for six different robots in both simulated and real-world environments. Experimental results demonstrate that PhysVLM outperforms existing models, achieving a 14% improvement over GPT-4o on EQA-phys and surpassing advanced embodied VLMs such as RoboMamba and SpatialVLM on the RoboVQA-val and OpenEQA benchmarks. Additionally, the S-P Map shows strong compatibility with various VLMs, and its integration into GPT-4o-mini yields a 7.1% performance improvement.
Claim
Understanding the environment and a robot’s physical reachability is crucial for task execution.
Synthetic Data is an Elegant GIFT for Continual Vision-Language Models Paper
Abstract
Pre-trained Vision-Language Models (VLMs) require Continual Learning (CL) to efficiently update their knowledge and adapt to various downstream tasks without retraining from scratch. However, for VLMs, in addition to the loss of knowledge previously learned from downstream tasks, pre-training knowledge is also corrupted during continual fine-tuning. This issue is exacerbated by the unavailability of original pre-training data, leaving VLM’s generalization ability degrading. In this paper, we propose GIFT, a novel continual fine-tuning approach that utilizes synthetic data to overcome catastrophic forgetting in VLMs. Taking advantage of recent advances in text-to-image synthesis, we employ a pre-trained diffusion model to recreate both pre-training and learned downstream task data. In this way, the VLM can revisit previous knowledge through distillation on matching diffusion-generated images and corresponding text prompts. Leveraging the broad distribution and high alignment between synthetic image-text pairs in VLM’s feature space, we propose a contrastive distillation loss along with an image-text alignment constraint. To further combat in-distribution overfitting and enhance distillation performance with limited amount of generated data, we incorporate adaptive weight consolidation, utilizing Fisher information from these synthetic image-text pairs and achieving a better stability-plasticity balance. Extensive experiments demonstrate that our method consistently outperforms previous state-of-the-art approaches across various settings.
Claim
Pre-trained Vision-Language Models (VLMs) require Continual Learning (CL) to efficiently update their knowledge and adapt to various downstream tasks without retraining from scratch.
Identifying and Mitigating Position Bias of Multi-image Vision-Language Models Paper
Abstract
The evolution of Large Vision-Language Models (LVLMs) has progressed from single to multi-image reasoning. Despite this advancement, our findings indicate that LVLMs struggle to robustly utilize information across multiple images, with predictions significantly affected by the alteration of image positions. To further explore this issue, we introduce Position-wise Question Answering (PQA), a meticulously designed task to quantify reasoning capabilities at each position. Our analysis reveals a pronounced position bias in LVLMs: open-source models excel in reasoning with images positioned later but underperform with those in the middle or at the beginning, while proprietary models show improved comprehension for images at the beginning and end but struggle with those in the middle. Motivated by this, we propose SoFt Attention (SoFA), a simple, training-free approach that mitigates this bias by employing linear interpolation between inter-image causal attention and bidirectional counterparts. Experimental results demonstrate that SoFA reduces position bias and enhances the reasoning performance of existing LVLMs. The code will be available https://github.com/xytian1008/sofa.
Claim
The evolution of Large Vision-Language Models (LVLMs) has progressed from single to multi-image reasoning.
POPEN: Preference-Based Optimization and Ensemble for LVLM-Based Reasoning Segmentation Paper
Abstract
Existing LVLM-based reasoning segmentation methods often suffer from imprecise segmentation results and hallucinations in their text responses. This paper introduces POPEN, a novel framework designed to address these issues and achieve improved results. POPEN includes a preference-based optimization method to finetune the LVLM, aligning it more closely with human preferences and thereby generating better text responses and segmentation results. Additionally, POPEN introduces a preference-based ensemble method for inference, which integrates multiple outputs from the LVLM using a preference-score-based attention mechanism for refinement. To better adapt to the segmentation task, we incorporate several task-specific designs in our POPEN framework, including a new approach for collecting segmentation preference data with a curriculum learning mechanism, and a novel preference optimization loss to refine the segmentation capability of the LVLM. Experiments demonstrate that our method achieves state-of-the-art performance in reasoning segmentation, exhibiting minimal hallucination in text responses and the highest segmentation accuracy compared to previous advanced methods like LISA and PixelLM. Project page is here.
Claim
Existing LVLM-based reasoning segmentation methods often suffer from imprecise segmentation results and hallucinations in their text responses.
Multi-Layer Visual Feature Fusion in Multimodal LLMs: Methods, Analysis, and Best Practices Paper
Abstract
Multimodal Large Language Models (MLLMs) have made significant advancements in recent years, with visual features playing an increasingly critical role in enhancing model performance. However, the integration of multi-layer visual features in MLLMs remains underexplored, particularly with regard to optimal layer selection and fusion strategies. Existing methods often rely on arbitrary design choices, leading to suboptimal outcomes. In this paper, we systematically investigate two core aspects of multi-layer visual feature fusion: (1) selecting the most effective visual layers and (2) identifying the best fusion approach with the language model. Our experiments reveal that while combining visual features from multiple stages improves generalization, incorporating additional features from the same stage typically leads to diminished performance. Furthermore, we find that direct fusion of multi-layer visual features at the input stage consistently yields superior and more stable performance across various configurations. We make all our code publicly available: https://github.com/EIT-NLP/Layer_Select_Fuse_for_MLLM.
Claim
Multimodal Large Language Models (MLLMs) have made significant advancements in recent years, with visual features playing an increasingly critical role in enhancing model performance.
Generalized Few-shot 3D Point Cloud Segmentation with Vision-Language Model Paper
Abstract
Generalized few-shot 3D point cloud segmentation (GFS-PCS) adapts models to new classes with few support samples while retaining base class segmentation. Existing GFS-PCS methods enhance prototypes via interacting with support or query features but remain limited by sparse knowledge from few-shot samples. Meanwhile, 3D vision-language models (3D VLMs), generalizing across open-world novel classes, contain rich but noisy novel class knowledge. In this work, we introduce a GFS-PCS framework that synergizes dense but noisy pseudo-labels from 3D VLMs with precise yet sparse few-shot samples to maximize the strengths of both, named GFS-VL. Specifically, we present a prototype-guided pseudo-label selection to filter low-quality regions, followed by an adaptive infilling strategy that combines knowledge from pseudo-label contexts and few-shot samples to adaptively label the filtered, un-labeled areas. Additionally, we design a novel-base mix strategy to embed few-shot samples into training scenes, preserving essential context for improved novel class learning. Moreover, recognizing the limited diversity in current GFS-PCS benchmarks, we introduce two challenging benchmarks with diverse novel classes for comprehensive generalization evaluation. Experiments validate the effectiveness of our framework across models and datasets. Our approach and benchmarks provide a solid foundation for advancing GFS-PCS in the real world. The code is at here.
Claim
Generalized few-shot 3D point cloud segmentation (GFS-PCS) adapts models to new classes with few support samples while retaining base class segmentation.
SOLVE: Synergy of Language-Vision and End-to-End Networks for Autonomous Driving Paper
Abstract
The integration of Vision-Language Models (VLMs) into autonomous driving systems has shown promise in addressing key challenges such as learning complexity, interpretability, and common-sense reasoning. However, existing approaches often struggle with efficient integration and real-time decision-making due to computational demands. In this paper, we introduce SOLVE, an innovative framework that synergizes VLMs with end-to-end (E2E) models to enhance autonomous vehicle planning. Our approach emphasizes knowledge sharing at the feature level through a shared visual encoder, enabling comprehensive interaction between VLM and E2E components. We propose a Trajectory Chain-of-Thought (T-CoT) paradigm, which progressively refines trajectory predictions, reducing uncertainty and improving accuracy. By employing a temporal decoupling strategy, SOLVE achieves efficient cooperation by aligning high-quality VLM outputs with E2E real-time performance. Evaluated on the nuScenes dataset, our method demonstrates significant improvements in trajectory prediction accuracy, paving the way for more robust and reliable autonomous driving systems.
Claim
The integration of Vision-Language Models (VLMs) into autonomous driving systems has shown promise in addressing key challenges such as learning complexity, interpretability, and common-sense reasoning.
SegAgent: Exploring Pixel Understanding Capabilities in MLLMs by Imitating Human Annotator Trajectories Paper
Abstract
While MLLMs have demonstrated impressive image understanding capabilities, they still struggle with pixel-level comprehension, limiting their practical applications. Current evaluation tasks such as VQA and visual grounding remain too coarse to assess fine-grained pixel comprehension accurately. Although segmentation is foundational for pixel-level understanding, existing methods often require MLLMs to generate implicit tokens, decoded through external pixel decoders. This approach disrupts the MLLM’s text output space, potentially compromising language capabilities and reducing flexibility and extensibility while failing to reflect the model’s intrinsic pixel-level understanding. Thus, we introduce the Human-Like Mask Annotation Task (HLMAT), a new paradigm where MLLMs mimic human annotators using interactive segmentation tools. Modelling segmentation as a multi-step Markov Decision Process, HLMAT enables MLLMs to iteratively generate text-based click points, achieving high-quality masks without architectural changes or implicit tokens. Through this setup, we develop SegAgent, a model fine-tuned on human-like annotation trajectories, which achieves performance comparable to SoTA methods and supports additional tasks like mask refinement and annotation filtering. HLMAT provides a protocol for assessing fine-grained pixel understanding in MLLMs and introduces a vision-centric, multi-step decision-making task that facilitates the exploration of MLLMs’ visual reasoning abilities. Our adaptations of policy improvement method StaR and PRM guided tree search further enhance model robustness in complex segmentation tasks, laying a foundation for future advancements in fine-grained visual perception and multi-step decision-making for MLLMs. Code can be found at https://github.com/aim-uofa/SegAgent.
Claim
While MLLMs have demonstrated impressive image understanding capabilities, they still struggle with pixel-level comprehension, limiting their practical applications.
GUI-Xplore: Empowering Generalizable GUI Agents with One Exploration Paper
Abstract
GUI agents hold significant potential to enhance the experience and efficiency of human-device interaction. However, current methods face challenges in generalizing across applications (apps) and tasks, primarily due to two fundamental limitations in existing datasets. First, these datasets overlook developer-induced structural variations among apps, limiting the transferability of knowledge across diverse software environments. Second, many of them focus solely on navigation tasks, which restricts their capacity to represent comprehensive software architectures and complex user interactions. To address these challenges, we introduce GUI-Xplore, a dataset meticulously designed to enhance cross-application and cross-task generalization via an exploration-and-reasoning framework. GUI-Xplore integrates pre-recorded exploration videos providing contextual insights, alongside five hierarchically structured downstream tasks designed to comprehensively evaluate GUI agent capabilities. To fully exploit GUI-Xplore’s unique features, we propose Xplore-Agent, a GUI agent framework that combines Action-aware GUI Modeling with Graph-Guided Environment Reasoning. Further experiments indicate that Xplore-Agent achieves a 10% improvement over existing methods in unfamiliar environments, yet there remains significant potential for further enhancement towards truly generalizable GUI agents. 1
Claim
GUI agents hold significant potential to enhance the experience and efficiency of human-device interaction.
Conical Visual Concentration for Efficient Large Vision-Language Models Paper
Abstract
In large vision-language models (LVLMs), images serve as inputs that carry a wealth of information. As the idiom "A picture is worth a thousand words" implies, representing a single image in current LVLMs can require hundreds or even thousands of tokens, which thereby severely impacting the efficiency. Previous approaches have attempted to reduce the number of image tokens either before or within the early layers of LVLMs. However, these strategies inevitably result in the loss of crucial image information. To address this challenge, we conduct an empirical study revealing that all visual tokens are necessary for LVLMs in the shallow layers, and token redundancy progressively increases in the deeper layers. To this end, we propose PyramidDrop, a visual redundancy reduction strategy for LVLMs to boost their efficiency in both training and inference with neglectable performance loss. Specifically, we partition the LVLM into several stages and drop part of the image tokens at the end of each stage with a pre-defined ratio. The dropping is based on a lightweight similarity calculation with a negligible time overhead. Extensive experiments demonstrate that PyramidDrop can achieve over 40% training time reduction and 55% inference FLOPs acceleration on leading LVLMs like LLaVA-NeXT, maintaining comparable multi-modal performance. Besides, PyramidDrop can also serve as a plug-and-play strategy to accelerate inference in a free way, with better performance and lower inference cost than counterparts. Our code is available at https://github.com/Cooperx521/PyramidDrop.
Claim
In large vision-language models (LVLMs), images serve as inputs that carry a wealth of information.
Rethinking Few-Shot Adaptation of Vision-Language Models in Two Stages Paper
Abstract
An old-school recipe for training a classifier is to (i) learn a good feature extractor and (ii) optimize a linear layer atop. When only a handful of samples are available per category, as in Few-Shot Adaptation (FSA), data are insufficient to fit a large number of parameters, rendering the above impractical. This is especially true with large pre-trained Vision-Language Models (VLMs), which motivated successful research at the intersection of Parameter-Efficient Fine-tuning (PEFT) and FSA. In this work, we start by analyzing the learning dynamics of PEFT techniques when trained on few-shot data from only a subset of categories, referred to as the "base" classes. We show that such dynamics naturally splits into two distinct phases: (i) task-level feature extraction and (ii) specialization to the available concepts. To accommodate this dynamic, we then depart from prompt- or adapter-based methods and tackle FSA differently. Specifically, given a fixed computational budget, we split it to (i) learn a task-specific feature extractor via PEFT and (ii) train a linear classifier on top. We call this scheme Two-Stage Few-Shot Adaptation (2SFS). Differently from established methods, our scheme enables a novel form of selective inference at a category level, i.e., at test time, only novel categories are embedded by the adapted text encoder, while embeddings of base categories are available within the classifier. Results with fixed hyperparameters across two settings, three backbones, and eleven datasets, show that 2SFS matches or surpasses the state-of-the-art, while established methods degrade significantly across settings. .
Claim
An old-school recipe for training a classifier is to (i) learn a good feature extractor and (ii) optimize a linear layer atop.
ReasonGrounder: LVLM-Guided Hierarchical Feature Splatting for Open-Vocabulary 3D Visual Grounding and Reasoning Paper
Abstract
Open-vocabulary 3D visual grounding and reasoning aim to localize objects in a scene based on implicit language descriptions, even when they are occluded. This ability is crucial for tasks such as vision-language navigation and autonomous robotics. However, current methods struggle because they rely heavily on fine-tuning with 3D annotations and mask proposals, which limits their ability to handle diverse semantics and common knowledge required for effective reasoning. In this work, we propose ReasonGrounder, an LVLM-guided framework that uses hierarchical 3D feature Gaussian fields for adaptive grouping based on physical scale, enabling open-vocabulary 3D grounding and reasoning. ReasonGrounder interprets implicit instructions using large vision-language models (LVLM) and localizes occluded objects through 3D Gaussian splatting. By incorporating 2D segmentation masks from the SAM and multi-view CLIP embeddings, ReasonGrounder selects Gaussian groups based on object scale, enabling accurate localization through both explicit and implicit language understanding, even in novel, occluded views. We also contribute ReasoningGD, a new dataset containing over 10K scenes and 2 million annotations for evaluating open-vocabulary 3D grounding and amodal perception under occlusion. Experiments show that ReasonGrounder significantly improves 3D grounding accuracy in real-world scenarios.
Claim
Open-vocabulary 3D visual grounding and reasoning aim to localize objects in a scene based on implicit language descriptions, even when they are occluded.
S4-Driver: Scalable Self-Supervised Driving Multimodal Large Language Model with Spatio-Temporal Visual Representation Paper
Abstract
The latest advancements in multi-modal large language models (MLLMs) have spurred a strong renewed interest in end-to-end motion planning approaches for autonomous driving. Many end-to-end approaches rely on human annotations to learn intermediate perception and prediction tasks, while purely self-supervised approaches—which directly learn from sensor inputs to generate planning trajectories without human annotations—often underperform the state of the art. We observe a key gap in the input representation space: end-to-end approaches built on MLLMs are often pretrained with reasoning tasks in 2D image space rather than the native 3D space in which autonomous vehicles plan. To this end, we propose S4-Driver, a scalable self-supervised motion planning algorithm with spatio-temporal visual representation, based on the popular PaLI [9] multimodal large language model. S4-Driver uses a novel sparse volume strategy to seamlessly transform the strong visual representation of MLLMs from perspective view to 3D space without the need to finetune the vision encoder. This representation aggregates multi-view and multi-frame visual inputs and enables better prediction of planning trajectories in 3D space. To validate our method, we run experiments on both nuScenes and Waymo Open Motion Dataset (with in-house camera data). Results show that S4-Driver performs favorably against existing supervised multi-task approaches while requiring no human annotations. It also demonstrates great scalability when pretrained on large volumes of unannotated driving logs.
Claim
The latest advancements in multi-modal large language models (MLLMs) have spurred a strong renewed interest in end-to-end motion planning approaches for autonomous driving.
MIMO: A medical vision language model with visual referring multimodal input and pixel grounding multimodal output Paper
Abstract
Currently, medical vision language models are widely used in medical vision question answering tasks. However, existing models are confronted with two issues: for input, the model only relies on text instructions and lacks direct understanding of visual clues in the image; for output, the model only gives text answers and lacks connection with key areas in the image. To address these issues, we propose a unified medical vision language model MIMO, with visual referring Multimodal Input and pixel grounding Multimodal Output. MIMO can not only combine visual clues and textual instructions to understand complex medical images and semantics, but can also ground medical terminologies in textual output within the image. To overcome the scarcity of relevant data in the medical field, we propose MIMOSeg, a comprehensive medical multimodal dataset including 895K samples. MIMOSeg is constructed from four different perspectives, covering basic instruction following and complex question answering with multimodal input and multimodal output. We conduct experiments on several downstream medical multimodal tasks. Extensive experimental results verify that MIMO can uniquely combine visual referring and pixel grounding capabilities, which are not available in previous models. Our project can be found in https://github.com/pkusixspace/MIMO.
Claim
Currently, medical vision language models are widely used in medical vision question answering tasks.
MASH-VLM: Mitigating Action-Scene Hallucination in Video-LLMs through Disentangled Spatial-Temporal Representations Paper
Abstract
In this work, we tackle action-scene hallucination in Video Large Language Models (Video-LLMs), where models incorrectly predict actions based on the scene context or scenes based on observed actions. We observe that existing Video-LLMs often suffer from action-scene hallucination due to two main factors. First, existing Video-LLMs intermingle spatial and temporal features by applying an attention operation across all tokens. Second, they use standard Rotary Position Embedding (RoPE), which causes the text tokens to overemphasize certain types of tokens depending on their sequential orders. To address these issues, we introduce MASH-VLM, Mitigating Action-Scene Hallucination in Video-LLMs through disentangled spatial-temporal representations. Our approach includes two key innovations: (1) DST-attention, a novel attention mechanism that disentangles spatial and temporal tokens within the LLM by using masked attention to restrict direct interactions be tween spatial and temporal tokens; (2) Harmonic-RoPE, which extends the dimensionality of the positional IDs, allowing spatial and temporal tokens to maintain balanced positions relative to the text tokens. To evaluate the action-scene hallucination in Video-LLMs, we introduce the UN-SCENE benchmark with 1,320 videos and 4,078 QA pairs. MASH-VLM achieves state-of-the-art performance on the UNSCENE benchmark, as well as on existing video understanding benchmarks.
Claim
In this work, we tackle action-scene hallucination in Video Large Language Models (Video-LLMs), where models incorrectly predict actions based on the scene context or scenes based on observed actions.
BlenderGym: Benchmarking Foundational Model Systems for Graphics Editing Paper
Abstract
3D graphics editing is crucial in applications like movie production and game design, yet it remains a time-consuming process that demands highly specialized domain expertise. Automating this process is challenging because graphical editing requires performing a variety of tasks, each requiring distinct skill sets. Recently, vision-language models (VLMs) have emerged as a powerful framework for automating the editing process, but their development and evaluation are bottlenecked by the lack of a comprehensive benchmark that requires human-level perception and presents real-world editing complexity. In this work, we present BlenderGym, the first comprehensive VLM system benchmark for 3D graphics editing. BlenderGym evaluates VLM systems through code-based 3D reconstruction tasks. We evaluate closed- and open-source VLM systems and observe that even the state-of-the-art VLM system struggles with tasks relatively easy for human Blender users. Enabled by BlenderGym, we study how inference scaling techniques impact VLM’s performance on graphics editing tasks. Notably, our findings reveal that the verifier used to guide the scaling of generation can itself be improved through inference scaling, complementing recent insights on inference scaling of LLM generation in coding and math tasks. We further show that inference compute is not uniformly effective and can be optimized by strategically distributing it between generation and verification.
Claim
3D graphics editing is crucial in applications like movie production and game design, yet it remains a time-consuming process that demands highly specialized domain expertise.
InteractVLM: 3D Interaction Reasoning from 2D Foundational Models Paper
Abstract
We introduce InteractVLM, a novel method to estimate 3D contact points on human bodies and objects from single in- the-wild images, enabling accurate human-object joint reconstruction in 3D. This is challenging due to occlusions, depth ambiguities, and widely varying object shapes. Existing methods rely on 3D contact annotations collected via expensive motion-capture systems or tedious manual labeling, limiting scalability and generalization. To overcome this, InteractVLM harnesses the broad visual knowledge of large Vision-Language Models (VLMs), fine-tuned with limited 3D contact data. However, directly applying these models is non-trivial, as they reason "only" in 2D, while human-object contact is inherently 3D. Thus we introduce a novel "Render-Localize-Lift" module that: (1) embeds 3D body and object surfaces in 2D space via multiview rendering, (2) trains a novel multi-view localization model (MV-Loc) to infer contacts in 2D, and (3) lifts these to 3D. Additionally, we propose a new task called Semantic Human Contact estimation, where human contact predictions are conditioned explicitly on object semantics, enabling richer interaction modeling. InteractVLM outperforms existing work on contact estimation and also facilitates 3D reconstruction from an in-the-wild image. To estimate 3D human and object pose, we infer initial body and object meshes, then infer contacts on both of these via InteractVLM, and last exploit these for fitting the meshes to image evidence. Results show that our approach performs promisingly in the wild. Code and models are available at https://interactvlm.is.tue.mpg.de.
Claim
We introduce InteractVLM, a novel method to estimate 3D contact points on human bodies and objects from single in- the-wild images, enabling accurate human-object joint reconstruction in 3D.
Building a Mind Palace: Structuring Environment-Grounded Semantic Graphs for Effective Long Video Analysis with LLMs Paper
Abstract
Long-form video understanding with Large Vision Language Models is challenged by the need to analyze temporally dispersed yet spatially concentrated key moments within limited context windows. In this work, we introduce VideoMindPalace, a new framework inspired by the "Mind Palace", which organizes critical video moments into a topologically structured semantic graph. VideoMindPalace organizes key information through (i) hand-object tracking and interaction, (ii) clustered activity zones representing specific areas of recurring activities, and (iii) environment layout mapping, allowing natural language parsing by LLMs to provide grounded insights on spatio-temporal and 3D context. In addition, we propose the Video Mind-Palace Benchmark (VMB), to assess human-like reasoning, including spatial localization, temporal reasoning, and layout-aware sequential understanding. Evaluated on VMB and established video QA datasets, including EgoSchema, NExT-QA, IntentQA, and the Active Memories Benchmark, VideoMindPalace demonstrates notable gains in spatiotemporal coherence and human-aligned reasoning, advancing long-form video analysis capabilities in VLMs. Project website: https://oodbag.github.io/VMP/.
Claim
Long-form video understanding with Large Vision Language Models is challenged by the need to analyze temporally dispersed yet spatially concentrated key moments within limited context windows.
Lifting the Veil on Visual Information Flow in MLLMs: Unlocking Pathways to Faster Inference Paper
Abstract
Multimodal large language models (MLLMs) improve performance on vision-language tasks by integrating visual features from pre-trained vision encoders into large language models (LLMs). However, how MLLMs process and utilize visual information remains unclear. In this paper, a shift in the dominant flow of visual information is uncovered: (1) in shallow layers, strong interactions are observed between image tokens and instruction tokens, where most visual information is injected into instruction tokens to form cross-modal semantic representations; (2) in deeper layers, image tokens primarily interact with each other, aggregating the remaining visual information to optimize semantic representations within visual modality. Based on these insights, we propose Hierarchical Modality-Aware Pruning (HiMAP), a plug-and-play inference acceleration method that dynamically prunes image tokens at specific layers, reducing computational costs by approximately 65% without sacrificing performance. Our findings offer a new understanding of visual information processing in MLLMs and provide a state-of-the-art solution for efficient inference. The code is available at https://github.com/ustc-hyin/HiMAP.
Claim
Multimodal large language models (MLLMs) improve performance on vision-language tasks by integrating visual features from pre-trained vision encoders into large language models (LLMs).
Stealthy Backdoor Attack in Self-Supervised Learning Vision Encoders for Large Vision Language Models Paper
Abstract
Self-supervised learning (SSL) vision encoders learn high-quality image representations and thus have become a vital part of developing vision modality of large vision language models (LVLMs). Due to the high cost of training such encoders, pre-trained encoders are widely shared and deployed into many LVLMs, which are security-critical or bear societal significance. Under this practical scenario, we reveal a new backdoor threat that significant visual hallucinations can be induced into these LVLMs by merely compromising vision encoders. Because of the sharing and reuse of these encoders, many downstream LVLMs may inherit backdoor behaviors from encoders, leading to widespread backdoors. In this work, we propose BadVision, the first method to exploit this vulnerability in SSL vision encoders for LVLMs with novel trigger optimization and backdoor learning techniques. We evaluate BadVision on two types of SSL encoders and LVLMs across eight benchmarks. We show that BadVision effectively drives the LVLMs to attacker-chosen hallucination with over 99% attack success rate, causing a 77.6% relative visual understanding error while maintaining the stealthiness. SoTA backdoor detection methods cannot detect our attack effectively. The code is available at https://github.com/6zHAOyi/BadVision.
Claim
Self-supervised learning (SSL) vision encoders learn high-quality image representations and thus have become a vital part of developing vision modality of large vision language models (LVLMs).
Bayesian Test-Time Adaptation for Vision-Language Models Paper
Abstract
Test-time adaptation with pre-trained vision-language models, such as CLIP, aims to adapt the model to new, potentially out-of-distribution test data. Existing methods calculate the similarity between visual embedding and learnable class embeddings, which are initialized by text embeddings, for zero-shot image classification. In this work, we first analyze this process based on Bayes theorem, and observe that the core factors influencing the final prediction are the likelihood and the prior. However, existing methods essentially focus on adapting class embeddings to adapt likelihood, but they often ignore the importance of prior. To address this gap, we propose a novel approach, Bayesian Class Adaptation (BCA), which in addition to continuously updating class embeddings to adapt likelihood, also uses the posterior of incoming samples to continuously update the prior for each class embedding. This dual updating mechanism allows the model to better adapt to distribution shifts and achieve higher prediction accuracy. Our method not only surpasses existing approaches in terms of performance metrics but also maintains superior inference rates and memory usage, making it highly efficient and practical for real-world applications.
Claim
Test-time adaptation with pre-trained vision-language models, such as CLIP, aims to adapt the model to new, potentially out-of-distribution test data.
DPC: Dual-Prompt Collaboration for Tuning Vision-Language Models Paper
Abstract
The Base-New Trade-off (BNT) problem universally exists during the optimization of CLIP-based prompt tuning, where continuous fine-tuning on base (target) classes leads to a simultaneous decrease of generalization ability on new (unseen) classes. Existing approaches attempt to regulate the prompt tuning process to balance BNT by appending constraints. However, imposed on the same target prompt, these constraints fail to fully avert the mutual exclusivity between the optimization directions for base and new. As a novel solution to this challenge, we propose the plug-andplay Dual-Prompt Collaboration (DPC) framework, the first that decoupling the optimization processes of base and new tasks at the prompt level. Specifically, we clone a learnable parallel prompt based on the backbone prompt, and introduce a variable Weighting-Decoupling framework to independently control the optimization directions of dual prompts specific to base or new tasks, thus avoiding the conflict in generalization. Meanwhile, we propose a Dynamic Hard Negative Optimizer, utilizing dual prompts to construct a more challenging optimization task on base classes for enhancement. For interpretability, we prove the feature channel invariance of the prompt vector during the optimization process, providing theoretical support for the Weighting-Decoupling of DPC. Extensive experiments on multiple backbones demonstrate that DPC can significantly improve base performance without introducing any external knowledge beyond the base classes, while maintaining generalization to new classes. Code is available at: https://github.com/JREion/DPC.
Claim
The Base-New Trade-off (BNT) problem universally exists during the optimization of CLIP-based prompt tuning, where continuous fine-tuning on base (target) classes leads to a simultaneous decrease of generalization ability on new (unseen) classes.
HOIGen-1M: A Large-scale Dataset for Human-Object Interaction Video Generation Paper
Abstract
Text-to-video (T2V) generation has made tremendous progress in generating complicated scenes based on texts. However, human-object interaction (HOI) often cannot be precisely generated by current T2V models due to the lack of large-scale videos with accurate captions for HOI. To address this issue, we introduce HOIGen-1M, the first large-scale dataset for HOI Generation, consisting of over one million high-quality videos collected from diverse sources. In particular, to guarantee the high quality of videos, we first design an efficient framework to automatically curate HOI videos using the powerful multimodal large language models (MLLMs), and then the videos are further cleaned by human annotators. Moreover, to obtain accurate textual captions for HOI videos, we design a novel video description method based on a Mixture-of-Multimodal-Experts (MoME) strategy that not only generates expressive captions but also eliminates the hallucination by individual MLLM. Further-more, due to the lack of an evaluation framework for gen-erated HOI videos, we propose two new metrics to assess the quality of generated videos in a coarse-to-fine manner. Extensive experiments reveal that current T2V models struggle to generate high-quality HOI videos and confirm that our HOIGen-1M dataset is instrumental for improving HOI video generation.
Claim
Text-to-video (T2V) generation has made tremendous progress in generating complicated scenes based on texts.
ProKeR: A Kernel Perspective on Few-Shot Adaptation of Large Vision-Language Models Paper
Abstract
The growing popularity of Contrastive Language-Image Pretraining (CLIP) has led to its widespread application in various visual downstream tasks. To enhance CLIP’s effectiveness and versatility, efficient few-shot adaptation techniques have been widely adopted. Among these approaches, training-free methods, particularly caching methods exemplified by Tip-Adapter, have gained attention for their lightweight adaptation without the need for additional fine-tuning. In this paper, we revisit Tip-Adapter from a kernel perspective, showing that caching methods function as local adapters and are connected to a well-established kernel literature. Drawing on this insight, we offer a theoretical understanding of how these methods operate and suggest multiple avenues for enhancing the Tip-Adapter baseline. Notably, our analysis shows the importance of incorporating global information in local adapters. Therefore, we subsequently propose a global method that learns a proximal regularizer in a reproducing kernel Hilbert space (RKHS) using CLIP as a base learner. Our method, which we call ProKeR (Proximal Kernel ridge Regression), has a closed form solution and achieves state-of-the-art performances across 11 datasets in the standard few-shot adaptation benchmark. Code is available at https://ybendou.github.io/ProKeR/
Claim
The growing popularity of Contrastive Language-Image Pretraining (CLIP) has led to its widespread application in various visual downstream tasks.
LoRASculpt: Sculpting LoRA for Harmonizing General and Specialized Knowledge in Multimodal Large Language Models Paper
Abstract
While Multimodal Large Language Models (MLLMs) excel at generalizing across modalities and tasks, effectively adapting them to specific downstream tasks while simultaneously retaining both general and specialized knowledge remains challenging. Although Low-Rank Adaptation (LoRA) is widely used to efficiently acquire specialized knowledge in MLLMs, it introduces substantial harmful redundancy during visual instruction tuning, which exacerbates the forgetting of general knowledge and degrades downstream task performance. To address this issue, we propose LoRASculpt to eliminate harmful redundant parameters, thereby harmonizing general and specialized knowledge. Specifically, under theoretical guarantees, we introduce sparse updates into LoRA to discard redundant parameters effectively. Furthermore, we propose a Conflict Mitigation Regularizer to refine the update trajectory of LoRA, mitigating knowledge conflicts with the pretrained weights. Extensive experimental results demonstrate that even at very high degree of sparsity (≤ 5%), our method simultaneously enhances generalization and downstream task performance. This confirms that our approach effectively mitigates the catastrophic forgetting issue and further promotes knowledge harmonization in MLLMs.
Claim
While Multimodal Large Language Models (MLLMs) excel at generalizing across modalities and tasks, effectively adapting them to specific downstream tasks while simultaneously retaining both general and specialized knowledge remains challenging.