Recent progress in VLMs has demonstrated impressive capabilities across a variety of tasks in the natural image domain. Motivated by these advancements, the remote sensing community has begun to adopt VLMs for remote sensing vision-language tasks, including scene understanding, image captioning, and visual question answering. However, existing remote sensing VLMs typically rely on closed-set scene understanding and focus on generic scene descriptions, yet lack the ability to incorporate external knowledge. This limitation hinders their capacity for semantic reasoning over complex or context-dependent queries that involve domain-specific or world knowledge. To address these challenges, we first introduced a multimodal Remote Sensing World Knowledge (RSWK) dataset, which comprises high-resolution satellite imagery and detailed textual descriptions for 14,141 well-known landmarks from 175 countries, integrating both remote sensing domain knowledge and broader world knowledge. Building upon this dataset, we proposed a novel Remote Sensing Retrieval-Augmented Generation (RS-RAG) framework, which consists of two key components. The Multi-Modal Knowledge Vector Database Construction module encodes remote sensing imagery and associated textual knowledge into a unified vector space. The Knowledge Retrieval and Response Generation module retrieves and re-ranks relevant knowledge based on image and/or text queries, and incorporates the retrieved content into a knowledge-augmented prompt to guide the VLM in producing contextually grounded responses. We validated the effectiveness of our approach on three representative vision-language tasks, including image captioning, image classification, and visual question answering, where RS-RAG significantly outperformed state-of-the-art baselines.
Monocular 3D occupancy prediction, aiming to predict the occupancy and semantics within interesting regions of 3D scenes from only 2D images, has garnered increasing attention recently for its vital role in 3D scene understanding. Predicting the 3D occupancy of large-scale outdoor scenes from 2D images is ill-posed and resource-intensive. In this paper, we present \textbf{DGOcc}, a \textbf{D}epth-aware \textbf{G}lobal query-based network for monocular 3D \textbf{Occ}upancy prediction. We first explore prior depth maps to extract depth context features that provide explicit geometric information for the occupancy network. Then, in order to fully exploit the depth context features, we propose a Global Query-based (GQ) Module. The cooperation of attention mechanisms and scale-aware operations facilitates the feature interaction between images and 3D voxels. Moreover, a Hierarchical Supervision Strategy (HSS) is designed to avoid upsampling the high-dimension 3D voxel features to full resolution, which mitigates GPU memory utilization and time cost. Extensive experiments on SemanticKITTI and SSCBench-KITTI-360 datasets demonstrate that the proposed method achieves the best performance on monocular semantic occupancy prediction while reducing GPU and time overhead.
Open-set semantic mapping is crucial for open-world robots. Current mapping approaches either are limited by the depth range or only map beyond-range entities in constrained settings, where overall they fail to combine within-range and beyond-range observations. Furthermore, these methods make a trade-off between fine-grained semantics and efficiency. We introduce RayFronts, a unified representation that enables both dense and beyond-range efficient semantic mapping. RayFronts encodes task-agnostic open-set semantics to both in-range voxels and beyond-range rays encoded at map boundaries, empowering the robot to reduce search volumes significantly and make informed decisions both within & beyond sensory range, while running at 8.84 Hz on an Orin AGX. Benchmarking the within-range semantics shows that RayFronts's fine-grained image encoding provides 1.34x zero-shot 3D semantic segmentation performance while improving throughput by 16.5x. Traditionally, online mapping performance is entangled with other system components, complicating evaluation. We propose a planner-agnostic evaluation framework that captures the utility for online beyond-range search and exploration, and show RayFronts reduces search volume 2.2x more efficiently than the closest online baselines.
Foundation models have demonstrated strong reasoning and generalization capabilities in driving-related tasks, including scene understanding, planning, and control. However, they still face challenges in hallucinations, uncertainty, and long inference latency. While existing foundation models have general knowledge of avoiding collisions, they often lack transportation-specific safety knowledge. To overcome these limitations, we introduce LetsPi, a physics-informed, dual-phase, knowledge-driven framework for safe, human-like trajectory planning. To prevent hallucinations and minimize uncertainty, this hybrid framework integrates Large Language Model (LLM) reasoning with physics-informed social force dynamics. LetsPi leverages the LLM to analyze driving scenes and historical information, providing appropriate parameters and target destinations (goals) for the social force model, which then generates the future trajectory. Moreover, the dual-phase architecture balances reasoning and computational efficiency through its Memory Collection phase and Fast Inference phase. The Memory Collection phase leverages the physics-informed LLM to process and refine planning results through reasoning, reflection, and memory modules, storing safe, high-quality driving experiences in a memory bank. Surrogate safety measures and physics-informed prompt techniques are introduced to enhance the LLM's knowledge of transportation safety and physical force, respectively. The Fast Inference phase extracts similar driving experiences as few-shot examples for new scenarios, while simplifying input-output requirements to enable rapid trajectory planning without compromising safety. Extensive experiments using the HighD dataset demonstrate that LetsPi outperforms baseline models across five safety metrics.See PDF for project Github link.
Audio-visual event localization (AVEL) plays a critical role in multimodal scene understanding. While existing datasets for AVEL predominantly comprise landscape-oriented long videos with clean and simple audio context, short videos have become the primary format of online video content due to the the proliferation of smartphones. Short videos are characterized by portrait-oriented framing and layered audio compositions (e.g., overlapping sound effects, voiceovers, and music), which brings unique challenges unaddressed by conventional methods. To this end, we introduce AVE-PM, the first AVEL dataset specifically designed for portrait mode short videos, comprising 25,335 clips that span 86 fine-grained categories with frame-level annotations. Beyond dataset creation, our empirical analysis shows that state-of-the-art AVEL methods suffer an average 18.66% performance drop during cross-mode evaluation. Further analysis reveals two key challenges of different video formats: 1) spatial bias from portrait-oriented framing introduces distinct domain priors, and 2) noisy audio composition compromise the reliability of audio modality. To address these issues, we investigate optimal preprocessing recipes and the impact of background music for AVEL on portrait mode videos. Experiments show that these methods can still benefit from tailored preprocessing and specialized model design, thus achieving improved performance. This work provides both a foundational benchmark and actionable insights for advancing AVEL research in the era of mobile-centric video content. Dataset and code will be released.
Visual emotion analysis or recognition has gained considerable attention due to the growing interest in understanding how images can convey rich semantics and evoke emotions in human perception. However, visual emotion analysis poses distinctive challenges compared to traditional vision tasks, especially due to the intricate relationship between general visual features and the different affective states they evoke, known as the affective gap. Researchers have used deep representation learning methods to address this challenge of extracting generalized features from entire images. However, most existing methods overlook the importance of specific emotional attributes such as brightness, colorfulness, scene understanding, and facial expressions. Through this paper, we introduce A4Net, a deep representation network to bridge the affective gap by leveraging four key attributes: brightness (Attribute 1), colorfulness (Attribute 2), scene context (Attribute 3), and facial expressions (Attribute 4). By fusing and jointly training all aspects of attribute recognition and visual emotion analysis, A4Net aims to provide a better insight into emotional content in images. Experimental results show the effectiveness of A4Net, showcasing competitive performance compared to state-of-the-art methods across diverse visual emotion datasets. Furthermore, visualizations of activation maps generated by A4Net offer insights into its ability to generalize across different visual emotion datasets.
Recent advances in scene understanding benefit a lot from depth maps because of the 3D geometry information, especially in complex conditions (e.g., low light and overexposed). Existing approaches encode depth maps along with RGB images and perform feature fusion between them to enable more robust predictions. Taking into account that depth can be regarded as a geometry supplement for RGB images, a straightforward question arises: Do we really need to explicitly encode depth information with neural networks as done for RGB images? Based on this insight, in this paper, we investigate a new way to learn RGBD feature representations and present DFormerv2, a strong RGBD encoder that explicitly uses depth maps as geometry priors rather than encoding depth information with neural networks. Our goal is to extract the geometry clues from the depth and spatial distances among all the image patch tokens, which will then be used as geometry priors to allocate attention weights in self-attention. Extensive experiments demonstrate that DFormerv2 exhibits exceptional performance in various RGBD semantic segmentation benchmarks. Code is available at: https://github.com/VCIP-RGBD/DFormer.
Thermal imaging is crucial for scene understanding, particularly in low-light and nighttime conditions. However, collecting large thermal datasets is costly and labor-intensive due to the specialized equipment required for infrared image capture. To address this challenge, researchers have explored visible-to-thermal image translation. Most existing methods rely on Generative Adversarial Networks (GANs) or Diffusion Models (DMs), treating the task as a style transfer problem. As a result, these approaches attempt to learn both the modality distribution shift and underlying physical principles from limited training data. In this paper, we propose F-ViTA, a novel approach that leverages the general world knowledge embedded in foundation models to guide the diffusion process for improved translation. Specifically, we condition an InstructPix2Pix Diffusion Model with zero-shot masks and labels from foundation models such as SAM and Grounded DINO. This allows the model to learn meaningful correlations between scene objects and their thermal signatures in infrared imagery. Extensive experiments on five public datasets demonstrate that F-ViTA outperforms state-of-the-art (SOTA) methods. Furthermore, our model generalizes well to out-of-distribution (OOD) scenarios and can generate Long-Wave Infrared (LWIR), Mid-Wave Infrared (MWIR), and Near-Infrared (NIR) translations from the same visible image. Code: https://github.com/JayParanjape/F-ViTA/tree/master.
Robot vision has greatly benefited from advancements in multimodal fusion techniques and vision-language models (VLMs). We systematically review the applications of multimodal fusion in key robotic vision tasks, including semantic scene understanding, simultaneous localization and mapping (SLAM), 3D object detection, navigation and localization, and robot manipulation. We compare VLMs based on large language models (LLMs) with traditional multimodal fusion methods, analyzing their advantages, limitations, and synergies. Additionally, we conduct an in-depth analysis of commonly used datasets, evaluating their applicability and challenges in real-world robotic scenarios. Furthermore, we identify critical research challenges such as cross-modal alignment, efficient fusion strategies, real-time deployment, and domain adaptation, and propose future research directions, including self-supervised learning for robust multimodal representations, transformer-based fusion architectures, and scalable multimodal frameworks. Through a comprehensive review, comparative analysis, and forward-looking discussion, we provide a valuable reference for advancing multimodal perception and interaction in robotic vision. A comprehensive list of studies in this survey is available at https://github.com/Xiaofeng-Han-Res/MF-RV.
Zero-shot 4D segmentation and recognition of arbitrary objects in Lidar is crucial for embodied navigation, with applications ranging from streaming perception to semantic mapping and localization. However, the primary challenge in advancing research and developing generalized, versatile methods for spatio-temporal scene understanding in Lidar lies in the scarcity of datasets that provide the necessary diversity and scale of annotations.To overcome these challenges, we propose SAL-4D (Segment Anything in Lidar--4D), a method that utilizes multi-modal robotic sensor setups as a bridge to distill recent developments in Video Object Segmentation (VOS) in conjunction with off-the-shelf Vision-Language foundation models to Lidar. We utilize VOS models to pseudo-label tracklets in short video sequences, annotate these tracklets with sequence-level CLIP tokens, and lift them to the 4D Lidar space using calibrated multi-modal sensory setups to distill them to our SAL-4D model. Due to temporal consistent predictions, we outperform prior art in 3D Zero-Shot Lidar Panoptic Segmentation (LPS) over $5$ PQ, and unlock Zero-Shot 4D-LPS.
Unsupervised panoptic segmentation aims to partition an image into semantically meaningful regions and distinct object instances without training on manually annotated data. In contrast to prior work on unsupervised panoptic scene understanding, we eliminate the need for object-centric training data, enabling the unsupervised understanding of complex scenes. To that end, we present the first unsupervised panoptic method that directly trains on scene-centric imagery. In particular, we propose an approach to obtain high-resolution panoptic pseudo labels on complex scene-centric data, combining visual representations, depth, and motion cues. Utilizing both pseudo-label training and a panoptic self-training strategy yields a novel approach that accurately predicts panoptic segmentation of complex scenes without requiring any human annotations. Our approach significantly improves panoptic quality, e.g., surpassing the recent state of the art in unsupervised panoptic segmentation on Cityscapes by 9.4% points in PQ.
This paper proposes a multi-view collaborative matching strategy for reliable track construction in complex scenarios. We observe that the pairwise matching paradigms applied to image set matching often result in ambiguous estimation when the selected independent pairs exhibit significant occlusions or extreme viewpoint changes. This challenge primarily stems from the inherent uncertainty in interpreting intricate 3D structures based on limited two-view observations, as the 3D-to-2D projection leads to significant information loss. To address this, we introduce CoMatcher, a deep multi-view matcher to (i) leverage complementary context cues from different views to form a holistic 3D scene understanding and (ii) utilize cross-view projection consistency to infer a reliable global solution. Building on CoMatcher, we develop a groupwise framework that fully exploits cross-view relationships for large-scale matching tasks. Extensive experiments on various complex scenarios demonstrate the superiority of our method over the mainstream two-view matching paradigm.
The rapid development of Large Multimodal Models (LMMs) for 2D images and videos has spurred efforts to adapt these models for interpreting 3D scenes. However, the absence of large-scale 3D vision-language datasets has posed a significant obstacle. To address this issue, typical approaches focus on injecting 3D awareness into 2D LMMs by designing 3D input-level scene representations. This work provides a new perspective. We introduce reconstructive visual instruction tuning with 3D-awareness (Ross3D), which integrates 3D-aware visual supervision into the training procedure. Specifically, it incorporates cross-view and global-view reconstruction. The former requires reconstructing masked views by aggregating overlapping information from other views. The latter aims to aggregate information from all available views to recover Bird's-Eye-View images, contributing to a comprehensive overview of the entire scene. Empirically, Ross3D achieves state-of-the-art performance across various 3D scene understanding benchmarks. More importantly, our semi-supervised experiments demonstrate significant potential in leveraging large amounts of unlabeled 3D vision-only data.
As human-robot collaboration advances, natural and flexible communication methods are essential for effective robot control. Traditional methods relying on a single modality or rigid rules struggle with noisy or misaligned data as well as with object descriptions that do not perfectly fit the predefined object names (e.g. 'Pick that red object'). We introduce TransforMerger, a transformer-based reasoning model that infers a structured action command for robotic manipulation based on fused voice and gesture inputs. Our approach merges multimodal data into a single unified sentence, which is then processed by the language model. We employ probabilistic embeddings to handle uncertainty and we integrate contextual scene understanding to resolve ambiguous references (e.g., gestures pointing to multiple objects or vague verbal cues like "this"). We evaluate TransforMerger in simulated and real-world experiments, demonstrating its robustness to noise, misalignment, and missing information. Our results show that TransforMerger outperforms deterministic baselines, especially in scenarios requiring more contextual knowledge, enabling more robust and flexible human-robot communication. Code and datasets are available at: http://imitrob.ciirc.cvut.cz/publications/transformerger.
3D point cloud semantic segmentation (PCSS) is a cornerstone for environmental perception in robotic systems and autonomous driving, enabling precise scene understanding through point-wise classification. While unsupervised domain adaptation (UDA) mitigates label scarcity in PCSS, existing methods critically overlook the inherent vulnerability to real-world perturbations (e.g., snow, fog, rain) and adversarial distortions. This work first identifies two intrinsic limitations that undermine current PCSS-UDA robustness: (a) unsupervised features overlap from unaligned boundaries in shared-class regions and (b) feature structure erosion caused by domain-invariant learning that suppresses target-specific patterns. To address the proposed problems, we propose a tripartite framework consisting of: 1) a robustness evaluation model quantifying resilience against adversarial attack/corruption types through robustness metrics; 2) an invertible attention alignment module (IAAM) enabling bidirectional domain mapping while preserving discriminative structure via attention-guided overlap suppression; and 3) a contrastive memory bank with quality-aware contrastive learning that progressively refines pseudo-labels with feature quality for more discriminative representations. Extensive experiments on SynLiDAR-to-SemanticPOSS adaptation demonstrate a maximum mIoU improvement of 14.3\% under adversarial attack.
Predicting human behavior in shared environments is crucial for safe and efficient human-robot interaction. Traditional data-driven methods to that end are pre-trained on domain-specific datasets, activity types, and prediction horizons. In contrast, the recent breakthroughs in Large Language Models (LLMs) promise open-ended cross-domain generalization to describe various human activities and make predictions in any context. In particular, Multimodal LLMs (MLLMs) are able to integrate information from various sources, achieving more contextual awareness and improved scene understanding. The difficulty in applying general-purpose MLLMs directly for prediction stems from their limited capacity for processing large input sequences, sensitivity to prompt design, and expensive fine-tuning. In this paper, we present a systematic analysis of applying pre-trained MLLMs for context-aware human behavior prediction. To this end, we introduce a modular multimodal human activity prediction framework that allows us to benchmark various MLLMs, input variations, In-Context Learning (ICL), and autoregressive techniques. Our evaluation indicates that the best-performing framework configuration is able to reach 92.8% semantic similarity and 66.1% exact label accuracy in predicting human behaviors in the target frame.
Accurate 6D object pose estimation from images is a key problem in object-centric scene understanding, enabling applications in robotics, augmented reality, and scene reconstruction. Despite recent advances, existing methods often produce physically inconsistent pose estimates, hindering their deployment in real-world scenarios. We introduce PhysPose, a novel approach that integrates physical reasoning into pose estimation through a postprocessing optimization enforcing non-penetration and gravitational constraints. By leveraging scene geometry, PhysPose refines pose estimates to ensure physical plausibility. Our approach achieves state-of-the-art accuracy on the YCB-Video dataset from the BOP benchmark and improves over the state-of-the-art pose estimation methods on the HOPE-Video dataset. Furthermore, we demonstrate its impact in robotics by significantly improving success rates in a challenging pick-and-place task, highlighting the importance of physical consistency in real-world applications.
We present OpenDriveVLA, a Vision-Language Action (VLA) model designed for end-to-end autonomous driving. OpenDriveVLA builds upon open-source pre-trained large Vision-Language Models (VLMs) to generate reliable driving actions, conditioned on 3D environmental perception, ego vehicle states, and driver commands. To bridge the modality gap between driving visual representations and language embeddings, we propose a hierarchical vision-language alignment process, projecting both 2D and 3D structured visual tokens into a unified semantic space. Besides, OpenDriveVLA models the dynamic relationships between the ego vehicle, surrounding agents, and static road elements through an autoregressive agent-env-ego interaction process, ensuring both spatially and behaviorally informed trajectory planning. Extensive experiments on the nuScenes dataset demonstrate that OpenDriveVLA achieves state-of-the-art results across open-loop trajectory planning and driving-related question-answering tasks. Qualitative analyses further illustrate OpenDriveVLA's superior capability to follow high-level driving commands and robustly generate trajectories under challenging scenarios, highlighting its potential for next-generation end-to-end autonomous driving. We will release our code to facilitate further research in this domain.
The DeepSeek models have shown exceptional performance in general scene understanding, question-answering (QA), and text generation tasks, owing to their efficient training paradigm and strong reasoning capabilities. In this study, we investigate the dialogue capabilities of the DeepSeek model in robotic surgery scenarios, focusing on tasks such as Single Phrase QA, Visual QA, and Detailed Description. The Single Phrase QA tasks further include sub-tasks such as surgical instrument recognition, action understanding, and spatial position analysis. We conduct extensive evaluations using publicly available datasets, including EndoVis18 and CholecT50, along with their corresponding dialogue data. Our empirical study shows that, compared to existing general-purpose multimodal large language models, DeepSeek-VL2 performs better on complex understanding tasks in surgical scenes. Additionally, although DeepSeek-V3 is purely a language model, we find that when image tokens are directly inputted, the model demonstrates better performance on single-sentence QA tasks. However, overall, the DeepSeek models still fall short of meeting the clinical requirements for understanding surgical scenes. Under general prompts, DeepSeek models lack the ability to effectively analyze global surgical concepts and fail to provide detailed insights into surgical scenarios. Based on our observations, we argue that the DeepSeek models are not ready for vision-language tasks in surgical contexts without fine-tuning on surgery-specific datasets.
The global rise in the number of people with physical disabilities, in part due to improvements in post-trauma survivorship and longevity, has amplified the demand for advanced assistive technologies to improve mobility and independence. Autonomous assistive robots, such as smart wheelchairs, require robust capabilities in spatial segmentation and semantic recognition to navigate complex built environments effectively. Place segmentation involves delineating spatial regions like rooms or functional areas, while semantic recognition assigns semantic labels to these regions, enabling accurate localization to user-specific needs. Existing approaches often utilize deep learning; however, these close-vocabulary detection systems struggle to interpret intuitive and casual human instructions. Additionally, most existing methods ignore the uncertainty of the scene recognition problem, leading to low success rates, particularly in ambiguous and complex environments. To address these challenges, we propose an open-vocabulary scene semantic segmentation and detection pipeline leveraging Vision Language Models (VLMs) and Large Language Models (LLMs). Our approach follows a 'Segment Detect Select' framework for open-vocabulary scene classification, enabling adaptive and intuitive navigation for assistive robots in built environments.