Generalizable depth completion enables the acquisition of dense metric depth maps for unseen environments, offering robust perception capabilities for various downstream tasks. However, training such models typically requires large-scale datasets with metric depth labels, which are often labor-intensive to collect. This paper presents PacGDC, a label-efficient technique that enhances data diversity with minimal annotation effort for generalizable depth completion. PacGDC builds on novel insights into inherent ambiguities and consistencies in object shapes and positions during 2D-to-3D projection, allowing the synthesis of numerous pseudo geometries for the same visual scene. This process greatly broadens available geometries by manipulating scene scales of the corresponding depth maps. To leverage this property, we propose a new data synthesis pipeline that uses multiple depth foundation models as scale manipulators. These models robustly provide pseudo depth labels with varied scene scales, affecting both local objects and global layouts, while ensuring projection consistency that supports generalization. To further diversify geometries, we incorporate interpolation and relocation strategies, as well as unlabeled images, extending the data coverage beyond the individual use of foundation models. Extensive experiments show that PacGDC achieves remarkable generalizability across multiple benchmarks, excelling in diverse scene semantics/scales and depth sparsity/patterns under both zero-shot and few-shot settings. Code: https://github.com/Wang-xjtu/PacGDC.
Commercial RGB-D cameras often produce noisy, incomplete depth maps for non-Lambertian objects. Traditional depth completion methods struggle to generalize due to the limited diversity and scale of training data. Recent advances exploit visual priors from pre-trained text-to-image diffusion models to enhance generalization in dense prediction tasks. However, we find that biases arising from training-inference mismatches in the vanilla diffusion framework significantly impair depth completion performance. Additionally, the lack of distinct visual features in non-Lambertian regions further hinders precise prediction. To address these issues, we propose \textbf{DidSee}, a diffusion-based framework for depth completion on non-Lambertian objects. First, we integrate a rescaled noise scheduler enforcing a zero terminal signal-to-noise ratio to eliminate signal leakage bias. Second, we devise a noise-agnostic single-step training formulation to alleviate error accumulation caused by exposure bias and optimize the model with a task-specific loss. Finally, we incorporate a semantic enhancer that enables joint depth completion and semantic segmentation, distinguishing objects from backgrounds and yielding precise, fine-grained depth maps. DidSee achieves state-of-the-art performance on multiple benchmarks, demonstrates robust real-world generalization, and effectively improves downstream tasks such as category-level pose estimation and robotic grasping.Project page: https://wenzhoulyu.github.io/DidSee/
Transparent and reflective objects in everyday environments pose significant challenges for depth sensors due to their unique visual properties, such as specular reflections and light transmission. These characteristics often lead to incomplete or inaccurate depth estimation, which severely impacts downstream geometry-based vision tasks, including object recognition, scene reconstruction, and robotic manipulation. To address the issue of missing depth information in transparent and reflective objects, we propose DCIRNet, a novel multimodal depth completion network that effectively integrates RGB images and depth maps to enhance depth estimation quality. Our approach incorporates an innovative multimodal feature fusion module designed to extract complementary information between RGB images and incomplete depth maps. Furthermore, we introduce a multi-stage supervision and depth refinement strategy that progressively improves depth completion and effectively mitigates the issue of blurred object boundaries. We integrate our depth completion model into dexterous grasping frameworks and achieve a $44\%$ improvement in the grasp success rate for transparent and reflective objects. We conduct extensive experiments on public datasets, where DCIRNet demonstrates superior performance. The experimental results validate the effectiveness of our approach and confirm its strong generalization capability across various transparent and reflective objects.
Recent advancements in 3D robotic manipulation have improved grasping of everyday objects, but transparent and specular materials remain challenging due to depth sensing limitations. While several 3D reconstruction and depth completion approaches address these challenges, they suffer from setup complexity or limited observation information utilization. To address this, leveraging the power of single view 3D object reconstruction approaches, we propose a training free framework SR3D that enables robotic grasping of transparent and specular objects from a single view observation. Specifically, given single view RGB and depth images, SR3D first uses the external visual models to generate 3D reconstructed object mesh based on RGB image. Then, the key idea is to determine the 3D object's pose and scale to accurately localize the reconstructed object back into its original depth corrupted 3D scene. Therefore, we propose view matching and keypoint matching mechanisms,which leverage both the 2D and 3D's inherent semantic and geometric information in the observation to determine the object's 3D state within the scene, thereby reconstructing an accurate 3D depth map for effective grasp detection. Experiments in both simulation and real world show the reconstruction effectiveness of SR3D.
Transparent and reflective objects pose significant challenges for depth sensors, resulting in incomplete depth information that adversely affects downstream robotic perception and manipulation tasks. To address this issue, we propose HTMNet, a novel hybrid model integrating Transformer, CNN, and Mamba architectures. The encoder is constructed based on a dual-branch Transformer-CNN framework, while the multi-scale fusion module leverages a Transformer-Mamba architecture, which also serves as the foundation for the decoder design. We introduce a novel multimodal fusion module grounded in self-attention mechanisms and state space models, marking the first application of the Mamba architecture in the field of transparent object depth completion and revealing its promising potential. Additionally, we design an innovative multi-scale fusion module for the decoder that combines channel attention, spatial attention, and multi-scale feature extraction techniques to effectively integrate multi-scale features through a down-fusion strategy. Extensive evaluations on multiple public datasets demonstrate that our model achieves state-of-the-art(SOTA) performance, validating the effectiveness of our approach.
In recent years, deep learning-based Monocular Depth Estimation (MDE) models have been widely applied in fields such as autonomous driving and robotics. However, their vulnerability to backdoor attacks remains unexplored. To fill the gap in this area, we conduct a comprehensive investigation of backdoor attacks against MDE models. Typically, existing backdoor attack methods can not be applied to MDE models. This is because the label used in MDE is in the form of a depth map. To address this, we propose BadDepth, the first backdoor attack targeting MDE models. BadDepth overcomes this limitation by selectively manipulating the target object's depth using an image segmentation model and restoring the surrounding areas via depth completion, thereby generating poisoned datasets for object-level backdoor attacks. To improve robustness in physical world scenarios, we further introduce digital-to-physical augmentation to adapt to the domain gap between the physical world and the digital domain. Extensive experiments on multiple models validate the effectiveness of BadDepth in both the digital domain and the physical world, without being affected by environmental factors.
Depth completion in dynamic scenes poses significant challenges due to rapid ego-motion and object motion, which can severely degrade the quality of input modalities such as RGB images and LiDAR measurements. Conventional RGB-D sensors often struggle to align precisely and capture reliable depth under such conditions. In contrast, event cameras with their high temporal resolution and sensitivity to motion at the pixel level provide complementary cues that are %particularly beneficial in dynamic environments.To this end, we propose EventDC, the first event-driven depth completion framework. It consists of two key components: Event-Modulated Alignment (EMA) and Local Depth Filtering (LDF). Both modules adaptively learn the two fundamental components of convolution operations: offsets and weights conditioned on motion-sensitive event streams. In the encoder, EMA leverages events to modulate the sampling positions of RGB-D features to achieve pixel redistribution for improved alignment and fusion. In the decoder, LDF refines depth estimations around moving objects by learning motion-aware masks from events. Additionally, EventDC incorporates two loss terms to further benefit global alignment and enhance local depth recovery. Moreover, we establish the first benchmark for event-based depth completion comprising one real-world and two synthetic datasets to facilitate future research. Extensive experiments on this benchmark demonstrate the superiority of our EventDC.
This work presents Prior Depth Anything, a framework that combines incomplete but precise metric information in depth measurement with relative but complete geometric structures in depth prediction, generating accurate, dense, and detailed metric depth maps for any scene. To this end, we design a coarse-to-fine pipeline to progressively integrate the two complementary depth sources. First, we introduce pixel-level metric alignment and distance-aware weighting to pre-fill diverse metric priors by explicitly using depth prediction. It effectively narrows the domain gap between prior patterns, enhancing generalization across varying scenarios. Second, we develop a conditioned monocular depth estimation (MDE) model to refine the inherent noise of depth priors. By conditioning on the normalized pre-filled prior and prediction, the model further implicitly merges the two complementary depth sources. Our model showcases impressive zero-shot generalization across depth completion, super-resolution, and inpainting over 7 real-world datasets, matching or even surpassing previous task-specific methods. More importantly, it performs well on challenging, unseen mixed priors and enables test-time improvements by switching prediction models, providing a flexible accuracy-efficiency trade-off while evolving with advancements in MDE models.
Interactive 3D generation is gaining momentum and capturing extensive attention for its potential to create immersive virtual experiences. However, a critical challenge in current 3D generation technologies lies in achieving real-time interactivity. To address this issue, we introduce WonderTurbo, the first real-time interactive 3D scene generation framework capable of generating novel perspectives of 3D scenes within 0.72 seconds. Specifically, WonderTurbo accelerates both geometric and appearance modeling in 3D scene generation. In terms of geometry, we propose StepSplat, an innovative method that constructs efficient 3D geometric representations through dynamic updates, each taking only 0.26 seconds. Additionally, we design QuickDepth, a lightweight depth completion module that provides consistent depth input for StepSplat, further enhancing geometric accuracy. For appearance modeling, we develop FastPaint, a 2-steps diffusion model tailored for instant inpainting, which focuses on maintaining spatial appearance consistency. Experimental results demonstrate that WonderTurbo achieves a remarkable 15X speedup compared to baseline methods, while preserving excellent spatial consistency and delivering high-quality output.
Depth completion, which estimates dense depth from sparse LiDAR and RGB images, has demonstrated outstanding performance in well-lit conditions. However, due to the limitations of RGB sensors, existing methods often struggle to achieve reliable performance in harsh environments, such as heavy rain and low-light conditions. Furthermore, we observe that ground truth depth maps often suffer from large missing measurements in adverse weather conditions such as heavy rain, leading to insufficient supervision. In contrast, thermal cameras are known for providing clear and reliable visibility in such conditions, yet research on thermal-LiDAR depth completion remains underexplored. Moreover, the characteristics of thermal images, such as blurriness, low contrast, and noise, bring unclear depth boundary problems. To address these challenges, we first evaluate the feasibility and robustness of thermal-LiDAR depth completion across diverse lighting (eg., well-lit, low-light), weather (eg., clear-sky, rainy), and environment (eg., indoor, outdoor) conditions, by conducting extensive benchmarks on the MS$^2$ and ViViD datasets. In addition, we propose a framework that utilizes COntrastive learning and Pseudo-Supervision (COPS) to enhance depth boundary clarity and improve completion accuracy by leveraging a depth foundation model in two key ways. First, COPS enforces a depth-aware contrastive loss between different depth points by mining positive and negative samples using a monocular depth foundation model to sharpen depth boundaries. Second, it mitigates the issue of incomplete supervision from ground truth depth maps by leveraging foundation model predictions as dense depth priors. We also provide in-depth analyses of the key challenges in thermal-LiDAR depth completion to aid in understanding the task and encourage future research.
Depth enhancement, which uses RGB images as guidance to convert raw signals from dToF into high-precision, dense depth maps, is a critical task in computer vision. Although existing super-resolution-based methods show promising results on public datasets, they often rely on idealized assumptions like accurate region correspondences and reliable dToF inputs, overlooking calibration errors that cause misalignment and anomaly signals inherent to dToF imaging, limiting real-world applicability. To address these challenges, we propose a novel completion-based method, named DEPTHOR, featuring advances in both the training strategy and model architecture. First, we propose a method to simulate real-world dToF data from the accurate ground truth in synthetic datasets to enable noise-robust training. Second, we design a novel network that incorporates monocular depth estimation (MDE), leveraging global depth relationships and contextual information to improve prediction in challenging regions. On the ZJU-L5 dataset, our training strategy significantly enhances depth completion models, achieving results comparable to depth super-resolution methods, while our model achieves state-of-the-art results, improving Rel and RMSE by 27% and 18%, respectively. On a more challenging set of dToF samples we collected, our method outperforms SOTA methods on preliminary stereo-based GT, improving Rel and RMSE by 23% and 22%, respectively. Our Code is available at https://github.com/ShadowBbBb/Depthor
The depth completion task is a critical problem in autonomous driving, involving the generation of dense depth maps from sparse depth maps and RGB images. Most existing methods employ a spatial propagation network to iteratively refine the depth map after obtaining an initial dense depth. In this paper, we propose DenseFormer, a novel method that integrates the diffusion model into the depth completion task. By incorporating the denoising mechanism of the diffusion model, DenseFormer generates the dense depth map by progressively refining an initial random depth distribution through multiple iterations. We propose a feature extraction module that leverages a feature pyramid structure, along with multi-layer deformable attention, to effectively extract and integrate features from sparse depth maps and RGB images, which serve as the guiding condition for the diffusion process. Additionally, this paper presents a depth refinement module that applies multi-step iterative refinement across various ranges to the dense depth results generated by the diffusion process. The module utilizes image features enriched with multi-scale information and sparse depth input to further enhance the accuracy of the predicted depth map. Extensive experiments on the KITTI outdoor scene dataset demonstrate that DenseFormer outperforms classical depth completion methods.
Generalized metric depth understanding is critical for precise vision-guided robotics, which current state-of-the-art (SOTA) vision-encoders do not support. To address this, we propose Vanishing Depth, a self-supervised training approach that extends pretrained RGB encoders to incorporate and align metric depth into their feature embeddings. Based on our novel positional depth encoding, we enable stable depth density and depth distribution invariant feature extraction. We achieve performance improvements and SOTA results across a spectrum of relevant RGBD downstream tasks - without the necessity of finetuning the encoder. Most notably, we achieve 56.05 mIoU on SUN-RGBD segmentation, 88.3 RMSE on Void's depth completion, and 83.8 Top 1 accuracy on NYUv2 scene classification. In 6D-object pose estimation, we outperform our predecessors of DinoV2, EVA-02, and Omnivore and achieve SOTA results for non-finetuned encoders in several related RGBD downstream tasks.
We present Pow3r, a novel large 3D vision regression model that is highly versatile in the input modalities it accepts. Unlike previous feed-forward models that lack any mechanism to exploit known camera or scene priors at test time, Pow3r incorporates any combination of auxiliary information such as intrinsics, relative pose, dense or sparse depth, alongside input images, within a single network. Building upon the recent DUSt3R paradigm, a transformer-based architecture that leverages powerful pre-training, our lightweight and versatile conditioning acts as additional guidance for the network to predict more accurate estimates when auxiliary information is available. During training we feed the model with random subsets of modalities at each iteration, which enables the model to operate under different levels of known priors at test time. This in turn opens up new capabilities, such as performing inference in native image resolution, or point-cloud completion. Our experiments on 3D reconstruction, depth completion, multi-view depth prediction, multi-view stereo, and multi-view pose estimation tasks yield state-of-the-art results and confirm the effectiveness of Pow3r at exploiting all available information. The project webpage is https://europe.naverlabs.com/pow3r.
Transparent and specular objects are frequently encountered in daily life, factories, and laboratories. However, due to the unique optical properties, the depth information on these objects is usually incomplete and inaccurate, which poses significant challenges for downstream robotics tasks. Therefore, it is crucial to accurately restore the depth information of transparent and specular objects. Previous depth completion methods for these objects usually use RGB information as an additional channel of the depth image to perform depth prediction. Due to the poor-texture characteristics of transparent and specular objects, these methods that rely heavily on color information tend to generate structure-less depth predictions. Moreover, these 2D methods cannot effectively explore the 3D structure hidden in the depth channel, resulting in depth ambiguity. To this end, we propose a geometry-aware assisted depth completion method for transparent and specular objects, which focuses on exploring the 3D structural cues of the scene. Specifically, besides extracting 2D features from RGB-D input, we back-project the input depth to a point cloud and build the 3D branch to extract hierarchical scene-level 3D structural features. To exploit 3D geometric information, we design several gated cross-modal fusion modules to effectively propagate multi-level 3D geometric features to the image branch. In addition, we propose an adaptive correlation aggregation strategy to appropriately assign 3D features to the corresponding 2D features. Extensive experiments on ClearGrasp, OOD, TransCG, and STD datasets show that our method outperforms other state-of-the-art methods. We further demonstrate that our method significantly enhances the performance of downstream robotic grasping tasks.
We present ProtoDepth, a novel prototype-based approach for continual learning of unsupervised depth completion, the multimodal 3D reconstruction task of predicting dense depth maps from RGB images and sparse point clouds. The unsupervised learning paradigm is well-suited for continual learning, as ground truth is not needed. However, when training on new non-stationary distributions, depth completion models will catastrophically forget previously learned information. We address forgetting by learning prototype sets that adapt the latent features of a frozen pretrained model to new domains. Since the original weights are not modified, ProtoDepth does not forget when test-time domain identity is known. To extend ProtoDepth to the challenging setting where the test-time domain identity is withheld, we propose to learn domain descriptors that enable the model to select the appropriate prototype set for inference. We evaluate ProtoDepth on benchmark dataset sequences, where we reduce forgetting compared to baselines by 52.2% for indoor and 53.2% for outdoor to achieve the state of the art.
Lightweight direct Time-of-Flight (dToF) sensors are ideal for 3D sensing on mobile devices. However, due to the manufacturing constraints of compact devices and the inherent physical principles of imaging, dToF depth maps are sparse and noisy. In this paper, we propose a novel video depth completion method, called SVDC, by fusing the sparse dToF data with the corresponding RGB guidance. Our method employs a multi-frame fusion scheme to mitigate the spatial ambiguity resulting from the sparse dToF imaging. Misalignment between consecutive frames during multi-frame fusion could cause blending between object edges and the background, which results in a loss of detail. To address this, we introduce an adaptive frequency selective fusion (AFSF) module, which automatically selects convolution kernel sizes to fuse multi-frame features. Our AFSF utilizes a channel-spatial enhancement attention (CSEA) module to enhance features and generates an attention map as fusion weights. The AFSF ensures edge detail recovery while suppressing high-frequency noise in smooth regions. To further enhance temporal consistency, We propose a cross-window consistency loss to ensure consistent predictions across different windows, effectively reducing flickering. Our proposed SVDC achieves optimal accuracy and consistency on the TartanAir and Dynamic Replica datasets. Code is available at https://github.com/Lan1eve/SVDC.
Transparent object manipulation remains a sig- nificant challenge in robotics due to the difficulty of acquiring accurate and dense depth measurements. Conventional depth sensors often fail with transparent objects, resulting in in- complete or erroneous depth data. Existing depth completion methods struggle with interframe consistency and incorrectly model transparent objects as Lambertian surfaces, leading to poor depth reconstruction. To address these challenges, we propose TranSplat, a surface embedding-guided 3D Gaussian Splatting method tailored for transparent objects. TranSplat uses a latent diffusion model to generate surface embeddings that provide consistent and continuous representations, making it robust to changes in viewpoint and lighting. By integrating these surface embeddings with input RGB images, TranSplat effectively captures the complexities of transparent surfaces, enhancing the splatting of 3D Gaussians and improving depth completion. Evaluations on synthetic and real-world transpar- ent object benchmarks, as well as robot grasping tasks, show that TranSplat achieves accurate and dense depth completion, demonstrating its effectiveness in practical applications. We open-source synthetic dataset and model: https://github. com/jeongyun0609/TranSplat
Depth completion endeavors to reconstruct a dense depth map from sparse depth measurements, leveraging the information provided by a corresponding color image. Existing approaches mostly hinge on single-scale propagation strategies that iteratively ameliorate initial coarse depth estimates through pixel-level message passing. Despite their commendable outcomes, these techniques are frequently hampered by computational inefficiencies and a limited grasp of scene context. To circumvent these challenges, we introduce LP-Net, an innovative framework that implements a multi-scale, progressive prediction paradigm based on Laplacian Pyramid decomposition. Diverging from propagation-based approaches, LP-Net initiates with a rudimentary, low-resolution depth prediction to encapsulate the global scene context, subsequently refining this through successive upsampling and the reinstatement of high-frequency details at incremental scales. We have developed two novel modules to bolster this strategy: 1) the Multi-path Feature Pyramid module, which segregates feature maps into discrete pathways, employing multi-scale transformations to amalgamate comprehensive spatial information, and 2) the Selective Depth Filtering module, which dynamically learns to apply both smoothness and sharpness filters to judiciously mitigate noise while accentuating intricate details. By integrating these advancements, LP-Net not only secures state-of-the-art (SOTA) performance across both outdoor and indoor benchmarks such as KITTI, NYUv2, and TOFDC, but also demonstrates superior computational efficiency. At the time of submission, LP-Net ranks 1st among all peer-reviewed methods on the official KITTI leaderboard.
Depth completion is a key task in autonomous driving, aiming to complete sparse LiDAR depth measurements into high-quality dense depth maps through image guidance. However, existing methods usually treat depth maps as an additional channel of color images, or directly perform convolution on sparse data, failing to fully exploit the 3D geometric information in depth maps, especially with limited performance in complex boundaries and sparse areas. To address these issues, this paper proposes a depth completion network combining channel attention mechanism and 3D global feature perception (CGA-Net). The main innovations include: 1) Utilizing PointNet++ to extract global 3D geometric features from sparse depth maps, enhancing the scene perception ability of low-line LiDAR data; 2) Designing a channel-attention-based multimodal feature fusion module to efficiently integrate sparse depth, RGB images, and 3D geometric features; 3) Combining residual learning with CSPN++ to optimize the depth refinement stage, further improving the completion quality in edge areas and complex scenes. Experiments on the KITTI depth completion dataset show that CGA-Net can significantly improve the prediction accuracy of dense depth maps, achieving a new state-of-the-art (SOTA), and demonstrating strong robustness to sparse and complex scenes.