Benjamin Busam

Guangyao Zhai, Dianye Huang, Shun-Cheng Wu et al.

6-DoF robotic grasping is a long-lasting but unsolved problem. Recent methods utilize strong 3D networks to extract geometric grasping representations from depth sensors, demonstrating superior accuracy on common objects but perform unsatisfactorily on photometrically challenging objects, e.g., objects in transparent or reflective materials. The bottleneck lies in that the surface of these objects can not reflect back accurate depth due to the absorption or refraction of light. In this paper, in contrast to exploiting the inaccurate depth data, we propose the first RGB-only 6-DoF grasping pipeline called MonoGraspNet that utilizes stable 2D features to simultaneously handle arbitrary object grasping and overcome the problems induced by photometrically challenging objects. MonoGraspNet leverages keypoint heatmap and normal map to recover the 6-DoF grasping poses represented by our novel representation parameterized with 2D keypoints with corresponding depth, grasping direction, grasping width, and angle. Extensive experiments in real scenes demonstrate that our method can achieve competitive results in grasping common objects and surpass the depth-based competitor by a large margin in grasping photometrically challenging objects. To further stimulate robotic manipulation research, we additionally annotate and open-source a multi-view and multi-scene real-world grasping dataset, containing 120 objects of mixed photometric complexity with 20M accurate grasping labels.

HyunJun Jung, Patrick Ruhkamp, Guangyao Zhai et al.

Learning-based methods to solve dense 3D vision problems typically train on 3D sensor data. The respectively used principle of measuring distances provides advantages and drawbacks. These are typically not compared nor discussed in the literature due to a lack of multi-modal datasets. Texture-less regions are problematic for structure from motion and stereo, reflective material poses issues for active sensing, and distances for translucent objects are intricate to measure with existing hardware. Training on inaccurate or corrupt data induces model bias and hampers generalisation capabilities. These effects remain unnoticed if the sensor measurement is considered as ground truth during the evaluation. This paper investigates the effect of sensor errors for the dense 3D vision tasks of depth estimation and reconstruction. We rigorously show the significant impact of sensor characteristics on the learned predictions and notice generalisation issues arising from various technologies in everyday household environments. For evaluation, we introduce a carefully designed dataset\footnote{dataset available at https://github.com/Junggy/HAMMER-dataset} comprising measurements from commodity sensors, namely D-ToF, I-ToF, passive/active stereo, and monocular RGB+P. Our study quantifies the considerable sensor noise impact and paves the way to improved dense vision estimates and targeted data fusion.

Guangyao Zhai, Yu Zheng, Ziwei Xu et al.

In this paper, we introduce DA$^2$, the first large-scale dual-arm dexterity-aware dataset for the generation of optimal bimanual grasping pairs for arbitrary large objects. The dataset contains about 9M pairs of parallel-jaw grasps, generated from more than 6000 objects and each labeled with various grasp dexterity measures. In addition, we propose an end-to-end dual-arm grasp evaluation model trained on the rendered scenes from this dataset. We utilize the evaluation model as our baseline to show the value of this novel and nontrivial dataset by both online analysis and real robot experiments. All data and related code will be open-sourced at https://sites.google.com/view/da2dataset.

Mahdi Saleh, Yige Wang, Nassir Navab et al.

Processing 3D data efficiently has always been a challenge. Spatial operations on large-scale point clouds, stored as sparse data, require extra cost. Attracted by the success of transformers, researchers are using multi-head attention for vision tasks. However, attention calculations in transformers come with quadratic complexity in the number of inputs and miss spatial intuition on sets like point clouds. We redesign set transformers in this work and incorporate them into a hierarchical framework for shape classification and part and scene segmentation. We propose our local attention unit, which captures features in a spatial neighborhood. We also compute efficient and dynamic global cross attentions by leveraging sampling and grouping at each iteration. Finally, to mitigate the non-heterogeneity of point clouds, we propose an efficient Multi-Scale Tokenization (MST), which extracts scale-invariant tokens for attention operations. The proposed hierarchical model achieves state-of-the-art shape classification in mean accuracy and yields results on par with the previous segmentation methods while requiring significantly fewer computations. Our proposed architecture predicts segmentation labels with around half the latency and parameter count of the previous most efficient method with comparable performance. The code is available at https://github.com/YigeWang-WHU/CloudAttention.

Magdalena Wysocki, Mohammad Farid Azampour, Christine Eilers et al.

We present a physics-enhanced implicit neural representation (INR) for ultrasound (US) imaging that learns tissue properties from overlapping US sweeps. Our proposed method leverages a ray-tracing-based neural rendering for novel view US synthesis. Recent publications demonstrated that INR models could encode a representation of a three-dimensional scene from a set of two-dimensional US frames. However, these models fail to consider the view-dependent changes in appearance and geometry intrinsic to US imaging. In our work, we discuss direction-dependent changes in the scene and show that a physics-inspired rendering improves the fidelity of US image synthesis. In particular, we demonstrate experimentally that our proposed method generates geometrically accurate B-mode images for regions with ambiguous representation owing to view-dependent differences of the US images. We conduct our experiments using simulated B-mode US sweeps of the liver and acquired US sweeps of a spine phantom tracked with a robotic arm. The experiments corroborate that our method generates US frames that enable consistent volume compounding from previously unseen views. To the best of our knowledge, the presented work is the first to address view-dependent US image synthesis using INR.

Yongzhi Su, Mahdi Saleh, Torben Fetzer et al.

Establishing correspondences from image to 3D has been a key task of 6DoF object pose estimation for a long time. To predict pose more accurately, deeply learned dense maps replaced sparse templates. Dense methods also improved pose estimation in the presence of occlusion. More recently researchers have shown improvements by learning object fragments as segmentation. In this work, we present a discrete descriptor, which can represent the object surface densely. By incorporating a hierarchical binary grouping, we can encode the object surface very efficiently. Moreover, we propose a coarse to fine training strategy, which enables fine-grained correspondence prediction. Finally, by matching predicted codes with object surface and using a PnP solver, we estimate the 6DoF pose. Results on the public LM-O and YCB-V datasets show major improvement over the state of the art w.r.t. ADD(-S) metric, even surpassing RGB-D based methods in some cases.

Chantal Pellegrini, Ege Özsoy, Benjamin Busam et al.

Conversational AI tools that can generate and discuss clinically correct radiology reports for a given medical image have the potential to transform radiology. Such a human-in-the-loop radiology assistant could facilitate a collaborative diagnostic process, thus saving time and improving the quality of reports. Towards this goal, we introduce RaDialog, the first thoroughly evaluated and publicly available large vision-language model for radiology report generation and interactive dialog. RaDialog effectively integrates visual image features and structured pathology findings with a large language model (LLM) while simultaneously adapting it to a specialized domain using parameter-efficient fine-tuning. To keep the conversational abilities of the underlying LLM, we propose a comprehensive, semi-automatically labeled, image-grounded instruct dataset for chest X-ray radiology tasks. By training with this dataset, our method achieves state-of-the-art clinical correctness in report generation and shows impressive abilities in interactive tasks such as correcting reports and answering questions, serving as a foundational step toward clinical dialog systems. Our code is available on github: https://github.com/ChantalMP/RaDialog.

Xue Hu, Xinghui Li, Benjamin Busam et al. · oxford

For visual manipulation tasks, we aim to represent image content with semantically meaningful features. However, learning implicit representations from images often lacks interpretability, especially when attributes are intertwined. We focus on the challenging task of extracting disentangled 3D attributes only from 2D image data. Specifically, we focus on human appearance and learn implicit pose, shape and garment representations of dressed humans from RGB images. Our method learns an embedding with disentangled latent representations of these three image properties and enables meaningful re-assembling of features and property control through a 2D-to-3D encoder-decoder structure. The 3D model is inferred solely from the feature map in the learned embedding space. To the best of our knowledge, our method is the first to achieve cross-domain disentanglement for this highly under-constrained problem. We qualitatively and quantitatively demonstrate our framework's ability to transfer pose, shape, and garments in 3D reconstruction on virtual data and show how an implicit shape loss can benefit the model's ability to recover fine-grained reconstruction details.

Hao Yu, Zheng Qin, Ji Hou et al.

The intrinsic rotation invariance lies at the core of matching point clouds with handcrafted descriptors. However, it is widely despised by recent deep matchers that obtain the rotation invariance extrinsically via data augmentation. As the finite number of augmented rotations can never span the continuous SO(3) space, these methods usually show instability when facing rotations that are rarely seen. To this end, we introduce RoITr, a Rotation-Invariant Transformer to cope with the pose variations in the point cloud matching task. We contribute both on the local and global levels. Starting from the local level, we introduce an attention mechanism embedded with Point Pair Feature (PPF)-based coordinates to describe the pose-invariant geometry, upon which a novel attention-based encoder-decoder architecture is constructed. We further propose a global transformer with rotation-invariant cross-frame spatial awareness learned by the self-attention mechanism, which significantly improves the feature distinctiveness and makes the model robust with respect to the low overlap. Experiments are conducted on both the rigid and non-rigid public benchmarks, where RoITr outperforms all the state-of-the-art models by a considerable margin in the low-overlapping scenarios. Especially when the rotations are enlarged on the challenging 3DLoMatch benchmark, RoITr surpasses the existing methods by at least 13 and 5 percentage points in terms of Inlier Ratio and Registration Recall, respectively.

Ivan Shugurov, Fu Li, Benjamin Busam et al.

We present a novel one-shot method for object detection and 6 DoF pose estimation, that does not require training on target objects. At test time, it takes as input a target image and a textured 3D query model. The core idea is to represent a 3D model with a number of 2D templates rendered from different viewpoints. This enables CNN-based direct dense feature extraction and matching. The object is first localized in 2D, then its approximate viewpoint is estimated, followed by dense 2D-3D correspondence prediction. The final pose is computed with PnP. We evaluate the method on LineMOD, Occlusion, Homebrewed, YCB-V and TLESS datasets and report very competitive performance in comparison to the state-of-the-art methods trained on synthetic data, even though our method is not trained on the object models used for testing.

Yamei Chen, Yan Di, Guangyao Zhai et al.

Category-level object pose estimation, aiming to predict the 6D pose and 3D size of objects from known categories, typically struggles with large intra-class shape variation. Existing works utilizing mean shapes often fall short of capturing this variation. To address this issue, we present SecondPose, a novel approach integrating object-specific geometric features with semantic category priors from DINOv2. Leveraging the advantage of DINOv2 in providing SE(3)-consistent semantic features, we hierarchically extract two types of SE(3)-invariant geometric features to further encapsulate local-to-global object-specific information. These geometric features are then point-aligned with DINOv2 features to establish a consistent object representation under SE(3) transformations, facilitating the mapping from camera space to the pre-defined canonical space, thus further enhancing pose estimation. Extensive experiments on NOCS-REAL275 demonstrate that SecondPose achieves a 12.4% leap forward over the state-of-the-art. Moreover, on a more complex dataset HouseCat6D which provides photometrically challenging objects, SecondPose still surpasses other competitors by a large margin.

Pengyuan Wang, HyunJun Jung, Yitong Li et al.

Object pose estimation is crucial for robotic applications and augmented reality. Beyond instance level 6D object pose estimation methods, estimating category-level pose and shape has become a promising trend. As such, a new research field needs to be supported by well-designed datasets. To provide a benchmark with high-quality ground truth annotations to the community, we introduce a multimodal dataset for category-level object pose estimation with photometrically challenging objects termed PhoCaL. PhoCaL comprises 60 high quality 3D models of household objects over 8 categories including highly reflective, transparent and symmetric objects. We developed a novel robot-supported multi-modal (RGB, depth, polarisation) data acquisition and annotation process. It ensures sub-millimeter accuracy of the pose for opaque textured, shiny and transparent objects, no motion blur and perfect camera synchronisation. To set a benchmark for our dataset, state-of-the-art RGB-D and monocular RGB methods are evaluated on the challenging scenes of PhoCaL.

HyunJun Jung, Guangyao Zhai, Shun-Cheng Wu et al.

Estimating 6D object poses is a major challenge in 3D computer vision. Building on successful instance-level approaches, research is shifting towards category-level pose estimation for practical applications. Current category-level datasets, however, fall short in annotation quality and pose variety. Addressing this, we introduce HouseCat6D, a new category-level 6D pose dataset. It features 1) multi-modality with Polarimetric RGB and Depth (RGBD+P), 2) encompasses 194 diverse objects across 10 household categories, including two photometrically challenging ones, and 3) provides high-quality pose annotations with an error range of only 1.35 mm to 1.74 mm. The dataset also includes 4) 41 large-scale scenes with comprehensive viewpoint and occlusion coverage, 5) a checkerboard-free environment, and 6) dense 6D parallel-jaw robotic grasp annotations. Additionally, we present benchmark results for leading category-level pose estimation networks.

Hanzhi Chen, Fabian Manhardt, Nassir Navab et al.

In this paper, we introduce neural texture learning for 6D object pose estimation from synthetic data and a few unlabelled real images. Our major contribution is a novel learning scheme which removes the drawbacks of previous works, namely the strong dependency on co-modalities or additional refinement. These have been previously necessary to provide training signals for convergence. We formulate such a scheme as two sub-optimisation problems on texture learning and pose learning. We separately learn to predict realistic texture of objects from real image collections and learn pose estimation from pixel-perfect synthetic data. Combining these two capabilities allows then to synthesise photorealistic novel views to supervise the pose estimator with accurate geometry. To alleviate pose noise and segmentation imperfection present during the texture learning phase, we propose a surfel-based adversarial training loss together with texture regularisation from synthetic data. We demonstrate that the proposed approach significantly outperforms the recent state-of-the-art methods without ground-truth pose annotations and demonstrates substantial generalisation improvements towards unseen scenes. Remarkably, our scheme improves the adopted pose estimators substantially even when initialised with much inferior performance.

Fu Li, Hao Yu, Ivan Shugurov et al.

Pose estimation of 3D objects in monocular images is a fundamental and long-standing problem in computer vision. Existing deep learning approaches for 6D pose estimation typically rely on the assumption of availability of 3D object models and 6D pose annotations. However, precise annotation of 6D poses in real data is intricate, time-consuming and not scalable, while synthetic data scales well but lacks realism. To avoid these problems, we present a weakly-supervised reconstruction-based pipeline, named NeRF-Pose, which needs only 2D object segmentation and known relative camera poses during training. Following the first-reconstruct-then-regress idea, we first reconstruct the objects from multiple views in the form of an implicit neural representation. Then, we train a pose regression network to predict pixel-wise 2D-3D correspondences between images and the reconstructed model. At inference, the approach only needs a single image as input. A NeRF-enabled PnP+RANSAC algorithm is used to estimate stable and accurate pose from the predicted correspondences. Experiments on LineMod and LineMod-Occlusion show that the proposed method has state-of-the-art accuracy in comparison to the best 6D pose estimation methods in spite of being trained only with weak labels. Besides, we extend the Homebrewed DB dataset with more real training images to support the weakly supervised task and achieve compelling results on this dataset. The extended dataset and code will be released soon.

Guangyao Zhai, Xiaoni Cai, Dianye Huang et al.

Object rearrangement is pivotal in robotic-environment interactions, representing a significant capability in embodied AI. In this paper, we present SG-Bot, a novel rearrangement framework that utilizes a coarse-to-fine scheme with a scene graph as the scene representation. Unlike previous methods that rely on either known goal priors or zero-shot large models, SG-Bot exemplifies lightweight, real-time, and user-controllable characteristics, seamlessly blending the consideration of commonsense knowledge with automatic generation capabilities. SG-Bot employs a three-fold procedure--observation, imagination, and execution--to adeptly address the task. Initially, objects are discerned and extracted from a cluttered scene during the observation. These objects are first coarsely organized and depicted within a scene graph, guided by either commonsense or user-defined criteria. Then, this scene graph subsequently informs a generative model, which forms a fine-grained goal scene considering the shape information from the initial scene and object semantics. Finally, for execution, the initial and envisioned goal scenes are matched to formulate robotic action policies. Experimental results demonstrate that SG-Bot outperforms competitors by a large margin.

Hao Yu, Ji Hou, Zheng Qin et al.

Successful point cloud registration relies on accurate correspondences established upon powerful descriptors. However, existing neural descriptors either leverage a rotation-variant backbone whose performance declines under large rotations, or encode local geometry that is less distinctive. To address this issue, we introduce RIGA to learn descriptors that are Rotation-Invariant by design and Globally-Aware. From the Point Pair Features (PPFs) of sparse local regions, rotation-invariant local geometry is encoded into geometric descriptors. Global awareness of 3D structures and geometric context is subsequently incorporated, both in a rotation-invariant fashion. More specifically, 3D structures of the whole frame are first represented by our global PPF signatures, from which structural descriptors are learned to help geometric descriptors sense the 3D world beyond local regions. Geometric context from the whole scene is then globally aggregated into descriptors. Finally, the description of sparse regions is interpolated to dense point descriptors, from which correspondences are extracted for registration. To validate our approach, we conduct extensive experiments on both object- and scene-level data. With large rotations, RIGA surpasses the state-of-the-art methods by a margin of 8\degree in terms of the Relative Rotation Error on ModelNet40 and improves the Feature Matching Recall by at least 5 percentage points on 3DLoMatch.

Yongzhi Su, Yan Di, Fabian Manhardt et al.

Despite monocular 3D object detection having recently made a significant leap forward thanks to the use of pre-trained depth estimators for pseudo-LiDAR recovery, such two-stage methods typically suffer from overfitting and are incapable of explicitly encapsulating the geometric relation between depth and object bounding box. To overcome this limitation, we instead propose OPA-3D, a single-stage, end-to-end, Occlusion-Aware Pixel-Wise Aggregation network that to jointly estimate dense scene depth with depth-bounding box residuals and object bounding boxes, allowing a two-stream detection of 3D objects, leading to significantly more robust detections. Thereby, the geometry stream denoted as the Geometry Stream, combines visible depth and depth-bounding box residuals to recover the object bounding box via explicit occlusion-aware optimization. In addition, a bounding box based geometry projection scheme is employed in an effort to enhance distance perception. The second stream, named as the Context Stream, directly regresses 3D object location and size. This novel two-stream representation further enables us to enforce cross-stream consistency terms which aligns the outputs of both streams, improving the overall performance. Extensive experiments on the public benchmark demonstrate that OPA-3D outperforms state-of-the-art methods on the main Car category, whilst keeping a real-time inference speed. We plan to release all codes and trained models soon.

Dekai Zhu, Guangyao Zhai, Yan Di et al.

Reliable multi-agent trajectory prediction is crucial for the safe planning and control of autonomous systems. Compared with single-agent cases, the major challenge in simultaneously processing multiple agents lies in modeling complex social interactions caused by various driving intentions and road conditions. Previous methods typically leverage graph-based message propagation or attention mechanism to encapsulate such interactions in the format of marginal probabilistic distributions. However, it is inherently sub-optimal. In this paper, we propose IPCC-TP, a novel relevance-aware module based on Incremental Pearson Correlation Coefficient to improve multi-agent interaction modeling. IPCC-TP learns pairwise joint Gaussian Distributions through the tightly-coupled estimation of the means and covariances according to interactive incremental movements. Our module can be conveniently embedded into existing multi-agent prediction methods to extend original motion distribution decoders. Extensive experiments on nuScenes and Argoverse 2 datasets demonstrate that IPCC-TP improves the performance of baselines by a large margin.

Shervin Dehghani, Michael Sommersperger, Peiyao Zhang et al.

In the last decade, various robotic platforms have been introduced that could support delicate retinal surgeries. Concurrently, to provide semantic understanding of the surgical area, recent advances have enabled microscope-integrated intraoperative Optical Coherent Tomography (iOCT) with high-resolution 3D imaging at near video rate. The combination of robotics and semantic understanding enables task autonomy in robotic retinal surgery, such as for subretinal injection. This procedure requires precise needle insertion for best treatment outcomes. However, merging robotic systems with iOCT introduces new challenges. These include, but are not limited to high demands on data processing rates and dynamic registration of these systems during the procedure. In this work, we propose a framework for autonomous robotic navigation for subretinal injection, based on intelligent real-time processing of iOCT volumes. Our method consists of an instrument pose estimation method, an online registration between the robotic and the iOCT system, and trajectory planning tailored for navigation to an injection target. We also introduce intelligent virtual B-scans, a volume slicing approach for rapid instrument pose estimation, which is enabled by Convolutional Neural Networks (CNNs). Our experiments on ex-vivo porcine eyes demonstrate the precision and repeatability of the method. Finally, we discuss identified challenges in this work and suggest potential solutions to further the development of such systems.

HyunJun Jung, Patrick Ruhkamp, Guangyao Zhai et al.

Depth estimation is a core task in 3D computer vision. Recent methods investigate the task of monocular depth trained with various depth sensor modalities. Every sensor has its advantages and drawbacks caused by the nature of estimates. In the literature, mostly mean average error of the depth is investigated and sensor capabilities are typically not discussed. Especially indoor environments, however, pose challenges for some devices. Textureless regions pose challenges for structure from motion, reflective materials are problematic for active sensing, and distances for translucent material are intricate to measure with existing sensors. This paper proposes HAMMER, a dataset comprising depth estimates from multiple commonly used sensors for indoor depth estimation, namely ToF, stereo, structured light together with monocular RGB+P data. We construct highly reliable ground truth depth maps with the help of 3D scanners and aligned renderings. A popular depth estimators is trained on this data and typical depth senosors. The estimates are extensively analyze on different scene structures. We notice generalization issues arising from various sensor technologies in household environments with challenging but everyday scene content. HAMMER, which we make publicly available, provides a reliable base to pave the way to targeted depth improvements and sensor fusion approaches.

Nicolas Schischka, Hannah Schieber, Mert Asim Karaoglu et al.

The accurate reconstruction of dynamic scenes with neural radiance fields is significantly dependent on the estimation of camera poses. Widely used structure-from-motion pipelines encounter difficulties in accurately tracking the camera trajectory when faced with separate dynamics of the scene content and the camera movement. To address this challenge, we propose Dynamic Motion-Aware Fast and Robust Camera Localization for Dynamic Neural Radiance Fields (DynaMoN). DynaMoN utilizes semantic segmentation and generic motion masks to handle dynamic content for initial camera pose estimation and statics-focused ray sampling for fast and accurate novel-view synthesis. Our novel iterative learning scheme switches between training the NeRF and updating the pose parameters for an improved reconstruction and trajectory estimation quality. The proposed pipeline shows significant acceleration of the training process. We extensively evaluate our approach on two real-world dynamic datasets, the TUM RGB-D dataset and the BONN RGB-D Dynamic dataset. DynaMoN improves over the state-of-the-art both in terms of reconstruction quality and trajectory accuracy. We plan to make our code public to enhance research in this area.

Lennart Bastian, Tony Danjun Wang, Tobias Czempiel et al.

Purpose: Recent advances in Surgical Data Science (SDS) have contributed to an increase in video recordings from hospital environments. While methods such as surgical workflow recognition show potential in increasing the quality of patient care, the quantity of video data has surpassed the scale at which images can be manually anonymized. Existing automated 2D anonymization methods under-perform in Operating Rooms (OR), due to occlusions and obstructions. We propose to anonymize multi-view OR recordings using 3D data from multiple camera streams. Methods: RGB and depth images from multiple cameras are fused into a 3D point cloud representation of the scene. We then detect each individual's face in 3D by regressing a parametric human mesh model onto detected 3D human keypoints and aligning the face mesh with the fused 3D point cloud. The mesh model is rendered into every acquired camera view, replacing each individual's face. Results: Our method shows promise in locating faces at a higher rate than existing approaches. DisguisOR produces geometrically consistent anonymizations for each camera view, enabling more realistic anonymization that is less detrimental to downstream tasks. Conclusion: Frequent obstructions and crowding in operating rooms leaves significant room for improvement for off-the-shelf anonymization methods. DisguisOR addresses privacy on a scene level and has the potential to facilitate further research in SDS.

Shiyu Li, Hannah Schieber, Kristoffer Waldow et al.

Multi-camera dynamic Augmented Reality (AR) applications require a camera pose estimation to leverage individual information from each camera in one common system. This can be achieved by combining contextual information, such as markers or objects, across multiple views. While commonly cameras are calibrated in an initial step or updated through the constant use of markers, another option is to leverage information already present in the scene, like known objects. Another downside of marker-based tracking is that markers have to be tracked inside the field-of-view (FoV) of the cameras. To overcome these limitations, we propose a constant dynamic camera pose estimation leveraging spatiotemporal FoV overlaps of known objects on the fly. To achieve that, we enhance the state-of-the-art object pose estimator to update our spatiotemporal scene graph, enabling a relation even among non-overlapping FoV cameras. To evaluate our approach, we introduce a multi-camera, multi-object pose estimation dataset with temporal FoV overlap, including static and dynamic cameras. Furthermore, in FoV overlapping scenarios, we outperform the state-of-the-art on the widely used YCB-V and T-LESS dataset in camera pose accuracy. Our performance on both previous and our proposed datasets validates the effectiveness of our marker-less approach for AR applications. The code and dataset are available on https://github.com/roth-hex-lab/IEEE-VR-2026-MultiCam.

Lennart Bastian, Alexander Baumann, Emily Hoppe et al.

Statistical shape models (SSMs) are an established way to represent the anatomy of a population with various clinically relevant applications. However, they typically require domain expertise, and labor-intensive landmark annotations to construct. We address these shortcomings by proposing an unsupervised method that leverages deep geometric features and functional correspondences to simultaneously learn local and global shape structures across population anatomies. Our pipeline significantly improves unsupervised correspondence estimation for SSMs compared to baseline methods, even on highly irregular surface topologies. We demonstrate this for two different anatomical structures: the thyroid and a multi-chamber heart dataset. Furthermore, our method is robust enough to learn from noisy neural network predictions, potentially enabling scaling SSMs to larger patient populations without manual segmentation annotation.

Stefano Gasperini, Alvaro Marcos-Ramiro, Michael Schmidt et al.

Panoptic segmentation methods assign a known class to each pixel given in input. Even for state-of-the-art approaches, this inevitably enforces decisions that systematically lead to wrong predictions for objects outside the training categories. However, robustness against out-of-distribution samples and corner cases is crucial in safety-critical settings to avoid dangerous consequences. Since real-world datasets cannot contain enough data points to adequately sample the long tail of the underlying distribution, models must be able to deal with unseen and unknown scenarios as well. Previous methods targeted this by re-identifying already-seen unlabeled objects. In this work, we propose the necessary step to extend segmentation with a new setting which we term holistic segmentation. Holistic segmentation aims to identify and separate objects of unseen, unknown categories into instances without any prior knowledge about them while performing panoptic segmentation of known classes. We tackle this new problem with U3HS, which finds unknowns as highly uncertain regions and clusters their corresponding instance-aware embeddings into individual objects. By doing so, for the first time in panoptic segmentation with unknown objects, our U3HS is trained without unknown categories, reducing assumptions and leaving the settings as unconstrained as in real-life scenarios. Extensive experiments on public data from MS COCO, Cityscapes, and Lost&Found demonstrate the effectiveness of U3HS for this new, challenging, and assumptions-free setting called holistic segmentation. Project page: https://holisticseg.github.io.

Lennart Bastian, Tobias Czempiel, Christian Heiliger et al.

The surgical operating room (OR) presents many opportunities for automation and optimization. Videos from various sources in the OR are becoming increasingly available. The medical community seeks to leverage this wealth of data to develop automated methods to advance interventional care, lower costs, and improve overall patient outcomes. Existing datasets from OR room cameras are thus far limited in size or modalities acquired, leaving it unclear which sensor modalities are best suited for tasks such as recognizing surgical action from videos. This study demonstrates that surgical action recognition performance can vary depending on the image modalities used. We perform a methodical analysis on several commonly available sensor modalities, presenting two fusion approaches that improve classification performance. The analyses are carried out on a set of multi-view RGB-D video recordings of 18 laparoscopic procedures.

Alexander Lehner, Stefano Gasperini, Alvaro Marcos-Ramiro et al.

Since real-world training datasets cannot properly sample the long tail of the underlying data distribution, corner cases and rare out-of-domain samples can severely hinder the performance of state-of-the-art models. This problem becomes even more severe for dense tasks, such as 3D semantic segmentation, where points of non-standard objects can be confidently associated to the wrong class. In this work, we focus on improving the generalization to out-of-domain data. We achieve this by augmenting the training set with adversarial examples. First, we learn a set of vectors that deform the objects in an adversarial fashion. To prevent the adversarial examples from being too far from the existing data distribution, we preserve their plausibility through a series of constraints, ensuring sensor-awareness and shapes smoothness. Then, we perform adversarial augmentation by applying the learned sample-independent vectors to the available objects when training a model. We conduct extensive experiments across a variety of scenarios on data from KITTI, Waymo, and CrashD for 3D object detection, and on data from SemanticKITTI, Waymo, and nuScenes for 3D semantic segmentation. Despite training on a standard single dataset, our approach substantially improves the robustness and generalization of both 3D object detection and 3D semantic segmentation methods to out-of-domain data.

Mert Asim Karaoglu, Viktoria Markova, Nassir Navab et al.

Unlike in natural images, in endoscopy there is no clear notion of an up-right camera orientation. Endoscopic videos therefore often contain large rotational motions, which require keypoint detection and description algorithms to be robust to these conditions. While most classical methods achieve rotation-equivariant detection and invariant description by design, many learning-based approaches learn to be robust only up to a certain degree. At the same time learning-based methods under moderate rotations often outperform classical approaches. In order to address this shortcoming, in this paper we propose RIDE, a learning-based method for rotation-equivariant detection and invariant description. Following recent advancements in group-equivariant learning, RIDE models rotation-equivariance implicitly within its architecture. Trained in a self-supervised manner on a large curation of endoscopic images, RIDE requires no manual labeling of training data. We test RIDE in the context of surgical tissue tracking on the SuPeR dataset as well as in the context of relative pose estimation on a repurposed version of the SCARED dataset. In addition we perform explicit studies showing its robustness to large rotations. Our comparison against recent learning-based and classical approaches shows that RIDE sets a new state-of-the-art performance on matching and relative pose estimation tasks and scores competitively on surgical tissue tracking.

Yan Di, Chenyangguang Zhang, Pengyuan Wang et al.

In this paper, we present a novel shape reconstruction method leveraging diffusion model to generate 3D sparse point cloud for the object captured in a single RGB image. Recent methods typically leverage global embedding or local projection-based features as the condition to guide the diffusion model. However, such strategies fail to consistently align the denoised point cloud with the given image, leading to unstable conditioning and inferior performance. In this paper, we present CCD-3DR, which exploits a novel centered diffusion probabilistic model for consistent local feature conditioning. We constrain the noise and sampled point cloud from the diffusion model into a subspace where the point cloud center remains unchanged during the forward diffusion process and reverse process. The stable point cloud center further serves as an anchor to align each point with its corresponding local projection-based features. Extensive experiments on synthetic benchmark ShapeNet-R2N2 demonstrate that CCD-3DR outperforms all competitors by a large margin, with over 40% improvement. We also provide results on real-world dataset Pix3D to thoroughly demonstrate the potential of CCD-3DR in real-world applications. Codes will be released soon

Ege Özsoy, Felix Holm, Mahdi Saleh et al.

Scene Graph Generation (SGG) is a visual understanding task, aiming to describe a scene as a graph of entities and their relationships with each other. Existing works rely on location labels in form of bounding boxes or segmentation masks, increasing annotation costs and limiting dataset expansion. Recognizing that many applications do not require location data, we break this dependency and introduce location-free scene graph generation (LF-SGG). This new task aims at predicting instances of entities, as well as their relationships, without the explicit calculation of their spatial localization. To objectively evaluate the task, the predicted and ground truth scene graphs need to be compared. We solve this NP-hard problem through an efficient branching algorithm. Additionally, we design the first LF-SGG method, Pix2SG, using autoregressive sequence modeling. We demonstrate the effectiveness of our method on three scene graph generation datasets as well as two downstream tasks, image retrieval and visual question answering, and show that our approach is competitive to existing methods while not relying on location cues.

HyunJun Jung, Patrick Ruhkamp, Nassir Navab et al.

This paper addresses the limitations of current datasets for 3D vision tasks in terms of accuracy, size, realism, and suitable imaging modalities for photometrically challenging objects. We propose a novel annotation and acquisition pipeline that enhances existing 3D perception and 6D object pose datasets. Our approach integrates robotic forward-kinematics, external infrared trackers, and improved calibration and annotation procedures. We present a multi-modal sensor rig, mounted on a robotic end-effector, and demonstrate how it is integrated into the creation of highly accurate datasets. Additionally, we introduce a freehand procedure for wider viewpoint coverage. Both approaches yield high-quality 3D data with accurate object and camera pose annotations. Our methods overcome the limitations of existing datasets and provide valuable resources for 3D vision research.

Shuhao Kang, Youqi Liao, Peijie Wang et al.

Text-to-point-cloud (T2P) localization aims to infer precise spatial positions within 3D point cloud maps from natural language descriptions, reflecting how humans perceive and communicate spatial layouts through language. However, existing methods largely rely on shallow text-point cloud correspondence without effective spatial reasoning, limiting their accuracy in complex environments. To address this limitation, we propose VLM-Loc, a framework that leverages the spatial reasoning capability of large vision-language models (VLMs) for T2P localization. Specifically, we transform point clouds into bird's-eye-view (BEV) images and scene graphs that jointly encode geometric and semantic context, providing structured inputs for the VLM to learn cross-modal representations bridging linguistic and spatial semantics. On top of these representations, we introduce a partial node assignment mechanism that explicitly associates textual cues with scene graph nodes, enabling interpretable spatial reasoning for accurate localization. To facilitate systematic evaluation across diverse scenes, we present CityLoc, a benchmark built from multi-source point clouds for fine-grained T2P localization. Experiments on CityLoc demonstrate VLM-Loc achieves superior accuracy and robustness compared to state-of-the-art methods. Our code, model, and dataset are available at \href{https://github.com/MCG-NKU/nku-3d-vision}{repository}.

Manoj Biswanath, Chenxin Cai, Hannah Schieber et al.

Efficient and robust 3D scene representation is crucial in autonomous driving, robotics, and related fields. While RGB images provide valuable content for 3D reconstruction, other modalities like thermal or depth can enable additional information on the environment. Lately, novel view synthesis methods like 3D Gaussian Splatting have started using multiple modalities to further boost their performance. But fusing or combining multimodal data can make the process slower and can bring in additional challenges. Therefore, our project aims to use single modality based on thermal infrared domain, by removing the reliance on visible light as much as possible. This single modality can be expected to be faster as it does not rely on multimodal data. We propose a method, Thermal-to-Depth Gaussian Splatting (TDg), that uses only thermal images and depth estimation in its architecture to derive the radiance fields. Our TDg method outperforms the MSMG (Multiple Single-Modal Gaussians) baseline in most cases on our test datasets, RGBT-Scenes and ThermalMix. On average, the rendering quality metrics such as learned perceptual image patch similarity (LPIPS), structural similarity index measure (SSIM), and peak signal-to-noise ratio (PSNR) of TDg are 1.12%, 0.034%, and 0.01% better than the baseline MSMG values. It also reduces the training time significantly, by 12 mins 47 secs (55% improvement). Overall, our method is successful in deriving these thermal radiance fields, which can ultimately have several applications, such as identifying heat sources critical in surveillance, search or rescue operations, and industrial inspections where temperature is widely used to monitor machines.

Yannick Verdié, Jifei Song, Barnabé Mas et al.

Learning-based depth estimation has witnessed recent progress in multiple directions; from self-supervision using monocular video to supervised methods offering highest accuracy. Complementary to supervision, further boosts to performance and robustness are gained by combining information from multiple signals. In this paper we systematically investigate key trade-offs associated with sensor and modality design choices as well as related model training strategies. Our study leads us to a new method, capable of connecting modality-specific advantages from polarisation, Time-of-Flight and structured-light inputs. We propose a novel pipeline capable of estimating depth from monocular polarisation for which we evaluate various training signals. The inversion of differentiable analytic models thereby connects scene geometry with polarisation and ToF signals and enables self-supervised and cross-modal learning. In the absence of existing multimodal datasets, we examine our approach with a custom-made multi-modal camera rig and collect CroMo; the first dataset to consist of synchronized stereo polarisation, indirect ToF and structured-light depth, captured at video rates. Extensive experiments on challenging video scenes confirm both qualitative and quantitative pipeline advantages where we are able to outperform competitive monocular depth estimation method.

Patrick Ruhkamp, Daoyi Gao, HyunJun Jung et al.

6D pose estimation pipelines that rely on RGB-only or RGB-D data show limitations for photometrically challenging objects with e.g. textureless surfaces, reflections or transparency. A supervised learning-based method utilising complementary polarisation information as input modality is proposed to overcome such limitations. This supervised approach is then extended to a self-supervised paradigm by leveraging physical characteristics of polarised light, thus eliminating the need for annotated real data. The methods achieve significant advancements in pose estimation by leveraging geometric information from polarised light and incorporating shape priors and invertible physical constraints.

Qilin Zhang, Jinyu Zhu, Olaf Wysocki et al.

Recent semantic 3D Gaussian Splatting (3DGS) methods primarily rely on 2D foundation models, often yielding ambiguous boundaries and limited support for structured urban semantics. While city models such as CityGML encode hierarchically organized semantics together with building geometry, these labels cannot be directly mapped to Gaussian primitives. We present GS4City, a hierarchical semantic Gaussian Splatting method that incorporates city-model priors for urban scene understanding. GS4City derives reliable image-aligned masks from Level of Detail (LoD) 3 CityGML models via two-pass raycasting, explicitly using parent-child relations to validate and recover fine-grained facade elements. It then fuses these geometry-grounded masks with foundation-model predictions to establish scene-consistent instance correspondences, and learns a compact identity encoding for each Gaussian under joint 2D identity supervision and 3D spatial regularization. Experiments on the TUM2TWIN and Gold Coast datasets show that GS4City effectively incorporates structured building semantics into Gaussian scene representations, outperforming existing 2D-driven semantic 3DGS baselines, including LangSplat and Gaga, by up to 15.8 IoU points in coarse building segmentation and 14.2 mIoU points in fine-grained semantic segmentation. By bridging structured city models and photorealistic Gaussian scene representations, GS4City enables semantically queryable and structure-aware urban reconstruction. Code is available at https://github.com/Jinyzzz/GS4City.

Alberto Rota, Mert Kiray, Mert Asim Karaoglu et al.

Specular highlights distort appearance, obscure texture, and hinder geometric reasoning in both natural and surgical imagery. We present UnReflectAnything, an RGB-only framework that removes highlights from a single image by predicting a highlight map together with a reflection-free diffuse reconstruction. The model uses a frozen vision transformer encoder to extract multi-scale features, a lightweight head to localize specular regions, and a token-level inpainting module that restores corrupted feature patches before producing the final diffuse image. To overcome the lack of paired supervision, we introduce a Virtual Highlight Synthesis pipeline that renders physically plausible specularities using monocular geometry, Fresnel-aware shading, and randomized lighting which enables training on arbitrary RGB images with correct geometric structure. UnReflectAnything generalizes across natural and surgical domains where non-Lambertian surfaces and non-uniform lighting create severe highlights and it achieves competitive performance with state-of-the-art results on several benchmarks. Project Page: https://alberto-rota.github.io/UnReflectAnything/

Shervin Dehghani, Benjamin Busam, Nassir Navab et al.

Despite its broad availability, volumetric information acquisition from Bright-Field Microscopy (BFM) is inherently difficult due to the projective nature of the acquisition process. We investigate the prediction of 3D cell instances from a set of BFM Z-Stack images. We propose a novel two-stage weakly supervised method for volumetric instance segmentation of cells which only requires approximate cell centroids annotation. Created pseudo-labels are thereby refined with a novel refinement loss with Z-stack guidance. The evaluations show that our approach can generalize not only to BFM Z-Stack data, but to other 3D cell imaging modalities. A comparison of our pipeline against fully supervised methods indicates that the significant gain in reduced data collection and labelling results in minor performance difference.

Ahmad AlMughrabi, Guillermo Rivo, Carlos Jiménez-Farfán et al.

Food image segmentation is a critical task for dietary analysis, enabling accurate estimation of food volume and nutrients. However, current methods suffer from limited multi-view data and poor generalization to new viewpoints. We introduce BenchSeg, a novel multi-view food video segmentation dataset and benchmark. BenchSeg aggregates 55 dish scenes (from Nutrition5k, Vegetables & Fruits, MetaFood3D, and FoodKit) with 25,284 meticulously annotated frames, capturing each dish under free 360° camera motion. We evaluate a diverse set of 20 state-of-the-art segmentation models (e.g., SAM-based, transformer, CNN, and large multimodal) on the existing FoodSeg103 dataset and evaluate them (alone and combined with video-memory modules) on BenchSeg. Quantitative and qualitative results demonstrate that while standard image segmenters degrade sharply under novel viewpoints, memory-augmented methods maintain temporal consistency across frames. Our best model based on a combination of SeTR-MLA+XMem2 outperforms prior work (e.g., improving over FoodMem by ~2.63% mAP), offering new insights into food segmentation and tracking for dietary analysis. We release BenchSeg to foster future research. The project page including the dataset annotations and the food segmentation models can be found at https://amughrabi.github.io/benchseg.

Shishir Reddy Vutukur, Rasmus Laurvig Haugaard, Junwen Huang et al.

Object pose distribution estimation is crucial in robotics for better path planning and handling of symmetric objects. Recent distribution estimation approaches employ contrastive learning-based approaches by maximizing the likelihood of a single pose estimate in the absence of a CAD model. We propose a pose distribution estimation method leveraging symmetry respecting correspondence distributions and shape information obtained using a CAD model. Contrastive learning-based approaches require an exhaustive amount of training images from different viewpoints to learn the distribution properly, which is not possible in realistic scenarios. Instead, we propose a pipeline that can leverage correspondence distributions and shape information from the CAD model, which are later used to learn pose distributions. Besides, having access to pose distribution based on correspondences before learning pose distributions conditioned on images, can help formulate the loss between distributions. The prior knowledge of distribution also helps the network to focus on getting sharper modes instead. With the CAD prior, our approach converges much faster and learns distribution better by focusing on learning sharper distribution near all the valid modes, unlike contrastive approaches, which focus on a single mode at a time. We achieve benchmark results on SYMSOL-I and T-Less datasets.

Liming Kuang, Yordanka Velikova, Mahdi Saleh et al.

Object pose estimation is a fundamental task in computer vision and robotics, yet most methods require extensive, dataset-specific training. Concurrently, large-scale vision language models show remarkable zero-shot capabilities. In this work, we bridge these two worlds by introducing ConceptPose, a framework for object pose estimation that is both training-free and model-free. ConceptPose leverages a vision-language-model (VLM) to create open-vocabulary 3D concept maps, where each point is tagged with a concept vector derived from saliency maps. By establishing robust 3D-3D correspondences across concept maps, our approach allows precise estimation of 6DoF relative pose. Without any object or dataset-specific training, our approach achieves state-of-the-art results on common zero shot relative pose estimation benchmarks, significantly outperforming existing methods by over 62% in ADD(-S) score, including those that utilize extensive dataset-specific training.

Guangyao Zhai, Evin Pınar Örnek, Dave Zhenyu Chen et al.

We present EchoScene, an interactive and controllable generative model that generates 3D indoor scenes on scene graphs. EchoScene leverages a dual-branch diffusion model that dynamically adapts to scene graphs. Existing methods struggle to handle scene graphs due to varying numbers of nodes, multiple edge combinations, and manipulator-induced node-edge operations. EchoScene overcomes this by associating each node with a denoising process and enables collaborative information exchange, enhancing controllable and consistent generation aware of global constraints. This is achieved through an information echo scheme in both shape and layout branches. At every denoising step, all processes share their denoising data with an information exchange unit that combines these updates using graph convolution. The scheme ensures that the denoising processes are influenced by a holistic understanding of the scene graph, facilitating the generation of globally coherent scenes. The resulting scenes can be manipulated during inference by editing the input scene graph and sampling the noise in the diffusion model. Extensive experiments validate our approach, which maintains scene controllability and surpasses previous methods in generation fidelity. Moreover, the generated scenes are of high quality and thus directly compatible with off-the-shelf texture generation. Code and trained models are open-sourced.

Weihang Li, Lorenzo Garattoni, Fabien Despinoy et al.

Learning model-free object pose estimation for unseen instances remains a fundamental challenge in 3D vision. Existing methods typically fall into two disjoint paradigms: category-level approaches predict absolute poses in a canonical space but rely on predefined taxonomies, while relative pose methods estimate cross-view transformations but cannot recover single-view absolute pose. In this work, we propose Object Pose Transformer (\ours{}), a unified feed-forward framework that bridges these paradigms through task factorization within a single model. \ours{} jointly predicts depth, point maps, camera parameters, and normalized object coordinates (NOCS) from RGB inputs, enabling both category-level absolute SA(3) pose and unseen-object relative SE(3) pose. Our approach leverages contrastive object-centric latent embeddings for canonicalization without requiring semantic labels at inference time, and uses point maps as a camera-space representation to enable multi-view relative geometric reasoning. Through cross-frame feature interaction and shared object embeddings, our model leverages relative geometric consistency across views to improve absolute pose estimation, reducing ambiguity in single-view predictions. Furthermore, \ours{} is camera-agnostic, learning camera intrinsics on-the-fly and supporting optional depth input for metric-scale recovery, while remaining fully functional in RGB-only settings. Extensive experiments on diverse benchmarks (NOCS, HouseCat6D, Omni6DPose, Toyota-Light) demonstrate state-of-the-art performance in both absolute and relative pose estimation tasks within a single unified architecture.

Amir Hamza, Davide Boscaini, Weihang Li et al.

Existing methods for instance-level 6D pose estimation typically rely on neural networks that either directly regress the pose in $\mathrm{SE}(3)$ or estimate it indirectly via local feature matching. The former struggle with object symmetries, while the latter fail in the absence of distinctive local features. To overcome these limitations, we propose a novel formulation of 6D pose estimation as a conditional flow matching problem in $\mathbb{R}^3$. We introduce Flose, a generative method that infers object poses via a denoising process conditioned on local features. While prior approaches based on conditional flow matching perform denoising solely based on geometric guidance, Flose integrates appearance-based semantic features to mitigate ambiguities caused by object symmetries. We further incorporate RANSAC-based registration to handle outliers. We validate Flose on five datasets from the established BOP benchmark. Flose outperforms prior methods with an average improvement of +4.5 Average Recall. Project Website : https://tev-fbk.github.io/Flose/

Adil Meric, Lin Geng Foo, Mert Kiray et al.

We present CoMoGen, a controllable video generation framework that generates realistic interactive dynamics from a single binary mask sequence conditioned on an input image. CoMoGen introduces a lightweight MaskAdapter that encodes binary mask sequences into a latent residual signal, injected into the Multi Modal Diffusion Transformer (MMDiT) model through a cosine-weighted schedule. Unlike the hierarchical coarse-to-fine design of UNet architectures, MMDiT operates as a sequence of uniform transformer blocks, making it difficult to identify which layers are responsible for the motion generation. Therefore, we propose a novel way to determine "Motion Layers" operating in the attention space of MMDiT. We fine-tune the model by using Low-Rank Adaptation (LoRA) to the Motion Layers, without requiring any architecture change in the MMDiT. This selective adaptation enables our method to focus on motion-critical components, yielding reduced computational cost. Despite its simplicity, CoMoGen enables precise subject motion and plausible interactions with surrounding humans, objects, and scenes. Comprehensive experiments on different datasets show that CoMoGen consistently outperforms prior controllable video generation methods and achieves state-of-the-art performance in motion fidelity and perceptual realism. Project page: mericadil.github.io/CoMoGen.

Shishir Reddy Vutukur, Mengkejiergeli Ba, Benjamin Busam et al.

In this paper, we propose a novel encoder-decoder architecture, named SABER, to learn the 6D pose of the object in the embedding space by learning shape representation at a given pose. This model enables us to learn pose by performing shape representation at a target pose from RGB image input. We perform shape representation as an auxiliary task which helps us in learning rotations space for an object based on 2D images. An image encoder predicts the rotation in the embedding space and the DeepSDF based decoder learns to represent the object's shape at the given pose. As our approach is shape based, the pipeline is suitable for any type of object irrespective of the symmetry. Moreover, we need only a CAD model of the objects to train SABER. Our pipeline is synthetic data based and can also handle symmetric objects without symmetry labels and, thus, no additional labeled training data is needed. The experimental evaluation shows that our method achieves close to benchmark results for both symmetric objects and asymmetric objects on Occlusion-LineMOD, and T-LESS datasets.

Ege Özsoy, Chantal Pellegrini, Tobias Czempiel et al.

Operating rooms (ORs) are complex, high-stakes environments requiring precise understanding of interactions among medical staff, tools, and equipment for enhancing surgical assistance, situational awareness, and patient safety. Current datasets fall short in scale, realism and do not capture the multimodal nature of OR scenes, limiting progress in OR modeling. To this end, we introduce MM-OR, a realistic and large-scale multimodal spatiotemporal OR dataset, and the first dataset to enable multimodal scene graph generation. MM-OR captures comprehensive OR scenes containing RGB-D data, detail views, audio, speech transcripts, robotic logs, and tracking data and is annotated with panoptic segmentations, semantic scene graphs, and downstream task labels. Further, we propose MM2SG, the first multimodal large vision-language model for scene graph generation, and through extensive experiments, demonstrate its ability to effectively leverage multimodal inputs. Together, MM-OR and MM2SG establish a new benchmark for holistic OR understanding, and open the path towards multimodal scene analysis in complex, high-stakes environments. Our code, and data is available at https://github.com/egeozsoy/MM-OR.

Luca Collorone, Mert Kiray, Indro Spinelli et al.

Realistic visual simulations are omnipresent, yet their creation requires computing time, rendering, and expert animation knowledge. Open-vocabulary visual effects generation from text inputs emerges as a promising solution that can unlock immense creative potential. However, current pipelines lack both physical realism and effective language interfaces, requiring slow offline optimization. In contrast, PhysTalk takes a 3D Gaussian Splatting (3DGS) scene as input and translates arbitrary user prompts into real time, physics based, interactive 4D animations. A large language model (LLM) generates executable code that directly modifies 3DGS parameters through lightweight proxies and particle dynamics. Notably, PhysTalk is the first framework to couple 3DGS directly with a physics simulator without relying on time consuming mesh extraction. While remaining open vocabulary, this design enables interactive 3D Gaussian animation via collision aware, physics based manipulation of arbitrary, multi material objects. Finally, PhysTalk is train-free and computationally lightweight: this makes 4D animation broadly accessible and shifts these workflows from a "render and wait" paradigm toward an interactive dialogue with a modern, physics-informed pipeline.

Ayaka Nanri, Klara Reichard, Mert Kiray et al.

Recent advances in 3D datasets and multimodal models have greatly improved natural language 3D scene understanding. However, most 3D referring segmentation methods do not explicitly represent the observer viewpoint, making spatial relations such as "left," "right," "front," and "behind" ambiguous and difficult to evaluate. We introduce a viewpoint-aware 3D referring segmentation dataset containing 220k benchmark samples, and scalable to tens of millions of viewpoint-conditioned samples through dense viewpoint sampling. In this dataset, target objects can only be identified through observer-centric spatial relations, making viewpoint-conditioned grounding necessary. We construct the benchmark by leveraging camera poses to automatically annotate observer-centric relations (left/right, front/behind) together with viewpoint-independent relations (above/under). Using this benchmark, we evaluate several existing 3D large multimodal models in a zero-shot setting and find that current models struggle with viewpoint-dependent spatial instructions. We further study how explicit viewpoint information can be incorporated into 3D large multimodal models. We introduce a viewpoint representation that encodes camera poses and conditions the model on the observation viewpoint, improving segmentation accuracy on viewpoint-dependent relations and increasing mIoU from 0.30 to 0.47 compared to a model without viewpoint conditioning. The dataset, code, and trained models will be made publicly available upon acceptance.