Fine-Grained Spatial and Verbal Losses for 3D Visual Grounding

3D visual grounding consists of identifying the instance in a 3D scene which is referred by an accompanying language description. While several architectures have been proposed within the commonly employed grounding-by-selection framework, the utilized losses are comparatively under-explored. In particular, most methods rely on a basic supervised cross-entropy loss on the predicted distribution over candidate instances, which fails to model both spatial relations between instances and the internal fine-grained word-level structure of the verbal referral. Sparse attempts to additionally supervise verbal embeddings globally by learning the class of the referred instance from the description or employing verbo-visual contrast to better separate instance embeddings do not fundamentally lift the aforementioned limitations. Responding to these shortcomings, we introduce two novel losses for 3D visual grounding: a visual-level offset loss on regressed vector offsets from each instance to the ground-truth referred instance and a language-related span loss on predictions for the word-level span of the referred instance in the description. In addition, we equip the verbo-visual fusion module of our new 3D visual grounding architecture AsphaltNet with a top-down bidirectional attentive fusion block, which enables the supervisory signals from our two losses to propagate to the respective converse branches of the network and thus aid the latter to learn context-aware instance embeddings and grounding-aware verbal embeddings. AsphaltNet proposes novel auxiliary losses to aid 3D visual grounding with competitive results compared to the state-of-the-art on the ReferIt3D benchmark.

Condition-Aware Multimodal Fusion for Robust Semantic Perception of Driving Scenes

Leveraging multiple sensors is crucial for robust semantic perception in autonomous driving, as each sensor type has complementary strengths and weaknesses. However, existing sensor fusion methods often treat sensors uniformly across all conditions, leading to suboptimal performance. By contrast, we propose a novel, condition-aware multimodal fusion approach for robust semantic perception of driving scenes. Our method, CAFuser uses an RGB camera input to classify environmental conditions and generate a Condition Token that guides the fusion of multiple sensor modalities. We further newly introduce modality-specific feature adapters to align diverse sensor inputs into a shared latent space, enabling efficient integration with a single and shared pre-trained backbone. By dynamically adapting sensor fusion based on the actual condition, our model significantly improves robustness and accuracy, especially in adverse-condition scenarios. We set the new state of the art with CAFuser on the MUSES dataset with 59.7 PQ for multimodal panoptic segmentation and 78.2 mIoU for semantic segmentation, ranking first on the public benchmarks.

Bayesian Self-Training for Semi-Supervised 3D Segmentation

3D segmentation is a core problem in computer vision and, similarly to many other dense prediction tasks, it requires large amounts of annotated data for adequate training. However, densely labeling 3D point clouds to employ fully-supervised training remains too labor intensive and expensive. Semi-supervised training provides a more practical alternative, where only a small set of labeled data is given, accompanied by a larger unlabeled set. This area thus studies the effective use of unlabeled data to reduce the performance gap that arises due to the lack of annotations. In this work, inspired by Bayesian deep learning, we first propose a Bayesian self-training framework for semi-supervised 3D semantic segmentation. Employing stochastic inference, we generate an initial set of pseudo-labels and then filter these based on estimated point-wise uncertainty. By constructing a heuristic $n$-partite matching algorithm, we extend the method to semi-supervised 3D instance segmentation, and finally, with the same building blocks, to dense 3D visual grounding. We demonstrate state-of-the-art results for our semi-supervised method on SemanticKITTI and ScribbleKITTI for 3D semantic segmentation and on ScanNet and S3DIS for 3D instance segmentation. We further achieve substantial improvements in dense 3D visual grounding over supervised-only baselines on ScanRefer. Our project page is available at ouenal.github.io/bst/.

Physically Feasible Semantic Segmentation

State-of-the-art semantic segmentation models are typically optimized in a data-driven fashion, minimizing solely per-pixel classification objectives on their training data. This purely data-driven paradigm often leads to absurd segmentations, especially when the domain of input images is shifted from the one encountered during training. For instance, state-of-the-art models may assign the label ``road'' to a segment which is located above a segment that is respectively labeled as ``sky'', although our knowledge of the physical world dictates that such a configuration is not feasible for images captured by forward-facing upright cameras. Our method, Physically Feasible Semantic Segmentation (PhyFea), extracts explicit physical constraints that govern spatial class relations from the training sets of semantic segmentation datasets and enforces a differentiable loss function that penalizes violations of these constraints to promote prediction feasibility. PhyFea yields significant performance improvements in mIoU over each state-of-the-art network we use as baseline across ADE20K, Cityscapes and ACDC, notably a $1.5\%$ improvement on ADE20K and a $2.1\%$ improvement on ACDC.

Sun Off, Lights On: Photorealistic Monocular Nighttime Simulation for Robust Semantic Perception

Nighttime scenes are hard to semantically perceive with learned models and annotate for humans. Thus, realistic synthetic nighttime data become all the more important for learning robust semantic perception at night, thanks to their accurate and cheap semantic annotations. However, existing data-driven or hand-crafted techniques for generating nighttime images from daytime counterparts suffer from poor realism. The reason is the complex interaction of highly spatially varying nighttime illumination, which differs drastically from its daytime counterpart, with objects of spatially varying materials in the scene, happening in 3D and being very hard to capture with such 2D approaches. The above 3D interaction and illumination shift have proven equally hard to model in the literature, as opposed to other conditions such as fog or rain. Our method, named Sun Off, Lights On (SOLO), is the first to perform nighttime simulation on single images in a photorealistic fashion by operating in 3D. It first explicitly estimates the 3D geometry, the materials and the locations of light sources of the scene from the input daytime image and relights the scene by probabilistically instantiating light sources in a way that accounts for their semantics and then running standard ray tracing. Not only is the visual quality and photorealism of our nighttime images superior to competing approaches including diffusion models, but the former images are also proven more beneficial for semantic nighttime segmentation in day-to-night adaptation. Code and data will be made publicly available.

TrafficBots V1.5: Traffic Simulation via Conditional VAEs and Transformers with Relative Pose Encoding

In this technical report we present TrafficBots V1.5, a baseline method for the closed-loop simulation of traffic agents. TrafficBots V1.5 achieves baseline-level performance and a 3rd place ranking in the Waymo Open Sim Agents Challenge (WOSAC) 2024. It is a simple baseline that combines TrafficBots, a CVAE-based multi-agent policy conditioned on each agent's individual destination and personality, and HPTR, the heterogeneous polyline transformer with relative pose encoding. To improve the performance on the WOSAC leaderboard, we apply scheduled teacher-forcing at the training time and we filter the sampled scenarios at the inference time. The code is available at https://github.com/zhejz/TrafficBotsV1.5.

Radar Fields: Frequency-Space Neural Scene Representations for FMCW Radar

Neural fields have been broadly investigated as scene representations for the reproduction and novel generation of diverse outdoor scenes, including those autonomous vehicles and robots must handle. While successful approaches for RGB and LiDAR data exist, neural reconstruction methods for radar as a sensing modality have been largely unexplored. Operating at millimeter wavelengths, radar sensors are robust to scattering in fog and rain, and, as such, offer a complementary modality to active and passive optical sensing techniques. Moreover, existing radar sensors are highly cost-effective and deployed broadly in robots and vehicles that operate outdoors. We introduce Radar Fields - a neural scene reconstruction method designed for active radar imagers. Our approach unites an explicit, physics-informed sensor model with an implicit neural geometry and reflectance model to directly synthesize raw radar measurements and extract scene occupancy. The proposed method does not rely on volume rendering. Instead, we learn fields in Fourier frequency space, supervised with raw radar data. We validate the effectiveness of the method across diverse outdoor scenarios, including urban scenes with dense vehicles and infrastructure, and in harsh weather scenarios, where mm-wavelength sensing is especially favorable.

UniDepth: Universal Monocular Metric Depth Estimation

Accurate monocular metric depth estimation (MMDE) is crucial to solving downstream tasks in 3D perception and modeling. However, the remarkable accuracy of recent MMDE methods is confined to their training domains. These methods fail to generalize to unseen domains even in the presence of moderate domain gaps, which hinders their practical applicability. We propose a new model, UniDepth, capable of reconstructing metric 3D scenes from solely single images across domains. Departing from the existing MMDE methods, UniDepth directly predicts metric 3D points from the input image at inference time without any additional information, striving for a universal and flexible MMDE solution. In particular, UniDepth implements a self-promptable camera module predicting dense camera representation to condition depth features. Our model exploits a pseudo-spherical output representation, which disentangles camera and depth representations. In addition, we propose a geometric invariance loss that promotes the invariance of camera-prompted depth features. Thorough evaluations on ten datasets in a zero-shot regime consistently demonstrate the superior performance of UniDepth, even when compared with methods directly trained on the testing domains. Code and models are available at: https://github.com/lpiccinelli-eth/unidepth

Vanishing-Point-Guided Video Semantic Segmentation of Driving Scenes

The estimation of implicit cross-frame correspondences and the high computational cost have long been major challenges in video semantic segmentation (VSS) for driving scenes. Prior works utilize keyframes, feature propagation, or cross-frame attention to address these issues. By contrast, we are the first to harness vanishing point (VP) priors for more effective segmentation. Intuitively, objects near VPs (i.e., away from the vehicle) are less discernible. Moreover, they tend to move radially away from the VP over time in the usual case of a forward-facing camera, a straight road, and linear forward motion of the vehicle. Our novel, efficient network for VSS, named VPSeg, incorporates two modules that utilize exactly this pair of static and dynamic VP priors: sparse-to-dense feature mining (DenseVP) and VP-guided motion fusion (MotionVP). MotionVP employs VP-guided motion estimation to establish explicit correspondences across frames and help attend to the most relevant features from neighboring frames, while DenseVP enhances weak dynamic features in distant regions around VPs. These modules operate within a context-detail framework, which separates contextual features from high-resolution local features at different input resolutions to reduce computational costs. Contextual and local features are integrated through contextualized motion attention (CMA) for the final prediction. Extensive experiments on two popular driving segmentation benchmarks, Cityscapes and ACDC, demonstrate that VPSeg outperforms previous SOTA methods, with only modest computational overhead.

MUSES: The Multi-Sensor Semantic Perception Dataset for Driving under Uncertainty

Achieving level-5 driving automation in autonomous vehicles necessitates a robust semantic visual perception system capable of parsing data from different sensors across diverse conditions. However, existing semantic perception datasets often lack important non-camera modalities typically used in autonomous vehicles, or they do not exploit such modalities to aid and improve semantic annotations in challenging conditions. To address this, we introduce MUSES, the MUlti-SEnsor Semantic perception dataset for driving in adverse conditions under increased uncertainty. MUSES includes synchronized multimodal recordings with 2D panoptic annotations for 2500 images captured under diverse weather and illumination. The dataset integrates a frame camera, a lidar, a radar, an event camera, and an IMU/GNSS sensor. Our new two-stage panoptic annotation protocol captures both class-level and instance-level uncertainty in the ground truth and enables the novel task of uncertainty-aware panoptic segmentation we introduce, along with standard semantic and panoptic segmentation. MUSES proves both effective for training and challenging for evaluating models under diverse visual conditions, and it opens new avenues for research in multimodal and uncertainty-aware dense semantic perception. Our dataset and benchmark will be made publicly available.