Monocular depth estimation (MDE) is a critical component of many medical tracking and mapping algorithms, particularly from endoscopic or laparoscopic video. However, because ground truth depth maps cannot be acquired from real patient data, supervised learning is not a viable approach to predict depth maps for medical scenes. Although self-supervised learning for MDE has recently gained attention, the outputs are difficult to evaluate reliably and each MDE's generalizability to other patients and anatomies is limited. This work evaluates the zero-shot performance of the newly released Depth Anything Model on medical endoscopic and laparoscopic scenes. We compare the accuracy and inference speeds of Depth Anything with other MDE models trained on general scenes as well as in-domain models trained on endoscopic data. Our findings show that although the zero-shot capability of Depth Anything is quite impressive, it is not necessarily better than other models in both speed and performance. We hope that this study can spark further research in employing foundation models for MDE in medical scenes.
In the realm of robot-assisted minimally invasive surgery, dynamic scene reconstruction can significantly enhance downstream tasks and improve surgical outcomes. Neural Radiance Fields (NeRF)-based methods have recently risen to prominence for their exceptional ability to reconstruct scenes but are hampered by slow inference speed, prolonged training, and inconsistent depth estimation. Some previous work utilizes ground truth depth for optimization but is hard to acquire in the surgical domain. To overcome these obstacles, we present Endo-4DGS, a real-time endoscopic dynamic reconstruction approach that utilizes 3D Gaussian Splatting (GS) for 3D representation. Specifically, we propose lightweight MLPs to capture temporal dynamics with Gaussian deformation fields. To obtain a satisfactory Gaussian Initialization, we exploit a powerful depth estimation foundation model, Depth-Anything, to generate pseudo-depth maps as a geometry prior. We additionally propose confidence-guided learning to tackle the ill-pose problems in monocular depth estimation and enhance the depth-guided reconstruction with surface normal constraints and depth regularization. Our approach has been validated on two surgical datasets, where it can effectively render in real-time, compute efficiently, and reconstruct with remarkable accuracy.
Depth completion is a crucial task in autonomous driving, aiming to convert a sparse depth map into a dense depth prediction. Due to its potentially rich semantic information, RGB image is commonly fused to enhance the completion effect. Image-guided depth completion involves three key challenges: 1) how to effectively fuse the two modalities; 2) how to better recover depth information; and 3) how to achieve real-time prediction for practical autonomous driving. To solve the above problems, we propose a concise but effective network, named CENet, to achieve high-performance depth completion with a simple and elegant structure. Firstly, we use a fast guidance module to fuse the two sensor features, utilizing abundant auxiliary features extracted from the color space. Unlike other commonly used complicated guidance modules, our approach is intuitive and low-cost. In addition, we find and analyze the optimization inconsistency problem for observed and unobserved positions, and a decoupled depth prediction head is proposed to alleviate the issue. The proposed decoupled head can better output the depth of valid and invalid positions with very few extra inference time. Based on the simple structure of dual-encoder and single-decoder, our CENet can achieve superior balance between accuracy and efficiency. In the KITTI depth completion benchmark, our CENet attains competitive performance and inference speed compared with the state-of-the-art methods. To validate the generalization of our method, we also evaluate on indoor NYUv2 dataset, and our CENet still achieve impressive results. The code of this work will be available at https://github.com/lmomoy/CHNet.
Methods for 3D reconstruction from posed frames require prior knowledge about the scene metric range, usually to recover matching cues along the epipolar lines and narrow the search range. However, such prior might not be directly available or estimated inaccurately in real scenarios -- e.g., outdoor 3D reconstruction from video sequences -- therefore heavily hampering performance. In this paper, we focus on multi-view depth estimation without requiring prior knowledge about the metric range of the scene by proposing RAMDepth, an efficient and purely 2D framework that reverses the depth estimation and matching steps order. Moreover, we demonstrate the capability of our framework to provide rich insights about the quality of the views used for prediction. Additional material can be found on our project page https://andreaconti.github.io/projects/range_agnostic_multi_view_depth.
Wide field-of-view (FoV) cameras efficiently capture large portions of the scene, which makes them attractive in multiple domains, such as automotive and robotics. For such applications, estimating depth from multiple images is a critical task, and therefore, a large amount of ground truth (GT) data is available. Unfortunately, most of the GT data is for pinhole cameras, making it impossible to properly train depth estimation models for large-FoV cameras. We propose the first method to train a stereo depth estimation model on the widely available pinhole data, and to generalize it to data captured with larger FoVs. Our intuition is simple: We warp the training data to a canonical, large-FoV representation and augment it to allow a single network to reason about diverse types of distortions that otherwise would prevent generalization. We show strong generalization ability of our approach on both indoor and outdoor datasets, which was not possible with previous methods.
Reconstructing deformable tissues from endoscopic videos is essential in many downstream surgical applications. However, existing methods suffer from slow rendering speed, greatly limiting their practical use. In this paper, we introduce EndoGaussian, a real-time endoscopic scene reconstruction framework built on 3D Gaussian Splatting (3DGS). By integrating the efficient Gaussian representation and highly-optimized rendering engine, our framework significantly boosts the rendering speed to a real-time level. To adapt 3DGS for endoscopic scenes, we propose two strategies, Holistic Gaussian Initialization (HGI) and Spatio-temporal Gaussian Tracking (SGT), to handle the non-trivial Gaussian initialization and tissue deformation problems, respectively. In HGI, we leverage recent depth estimation models to predict depth maps of input binocular/monocular image sequences, based on which pixels are re-projected and combined for holistic initialization. In SPT, we propose to model surface dynamics using a deformation field, which is composed of an efficient encoding voxel and a lightweight deformation decoder, allowing for Gaussian tracking with minor training and rendering burden. Experiments on public datasets demonstrate our efficacy against prior SOTAs in many aspects, including better rendering speed (195 FPS real-time, 100$\times$ gain), better rendering quality (37.848 PSNR), and less training overhead (within 2 min/scene), showing significant promise for intraoperative surgery applications. Code is available at: \url{https://yifliu3.github.io/EndoGaussian/}.
This paper introduces InverseMatrixVT3D, an efficient method for transforming multi-view image features into 3D feature volumes for 3D semantic occupancy prediction. Existing methods for constructing 3D volumes often rely on depth estimation, device-specific operators, or transformer queries, which hinders the widespread adoption of 3D occupancy models. In contrast, our approach leverages two projection matrices to store the static mapping relationships and matrix multiplications to efficiently generate global Bird's Eye View (BEV) features and local 3D feature volumes. Specifically, we achieve this by performing matrix multiplications between multi-view image feature maps and two sparse projection matrices. We introduce a sparse matrix handling technique for the projection matrices to optimize GPU memory usage. Moreover, a global-local attention fusion module is proposed to integrate the global BEV features with the local 3D feature volumes to obtain the final 3D volume. We also employ a multi-scale supervision mechanism to enhance performance further. Extensive experiments performed on the nuScenes and SemanticKITTI datasets reveal that our approach not only stands out for its simplicity and effectiveness but also achieves the top performance in detecting vulnerable road users (VRU), crucial for autonomous driving and road safety. The code has been made available at: https://github.com/DanielMing123/InverseMatrixVT3D
Sampling autonomy for icy moon lander missions requires understanding of topographic and photometric properties of the sampling terrain. Unavailability of high resolution visual datasets (either bird-eye view or point-of-view from a lander) is an obstacle for selection, verification or development of perception systems. We attempt to alleviate this problem by: 1) proposing Graphical Utility for Icy moon Surface Simulations (GUISS) framework, for versatile stereo dataset generation that spans the spectrum of bulk photometric properties, and 2) focusing on a stereo-based visual perception system and evaluating both traditional and deep learning-based algorithms for depth estimation from stereo matching. The surface reflectance properties of icy moon terrains (Enceladus and Europa) are inferred from multispectral datasets of previous missions. With procedural terrain generation and physically valid illumination sources, our framework can fit a wide range of hypotheses with respect to visual representations of icy moon terrains. This is followed by a study over the performance of stereo matching algorithms under different visual hypotheses. Finally, we emphasize the standing challenges to be addressed for simulating perception data assets for icy moons such as Enceladus and Europa. Our code can be found here: https://github.com/nasa-jpl/guiss.
In stereo-matching knowledge distillation methods of the self-supervised monocular depth estimation, the stereo-matching network's knowledge is distilled into a monocular depth network through pseudo-depth maps. In these methods, the learning-based stereo-confidence network is generally utilized to identify errors in the pseudo-depth maps to prevent transferring the errors. However, the learning-based stereo-confidence networks should be trained with ground truth (GT), which is not feasible in a self-supervised setting. In this paper, we propose a method to identify and filter errors in the pseudo-depth map using multiple disparity maps by checking their consistency without the need for GT and a training process. Experimental results show that the proposed method outperforms the previous methods and works well on various configurations by filtering out erroneous areas where the stereo-matching is vulnerable, especially such as textureless regions, occlusion boundaries, and reflective surfaces.
Recent advancements in learning-based Multi-View Stereo (MVS) methods have prominently featured transformer-based models with attention mechanisms. However, existing approaches have not thoroughly investigated the profound influence of transformers on different MVS modules, resulting in limited depth estimation capabilities. In this paper, we introduce MVSFormer++, a method that prudently maximizes the inherent characteristics of attention to enhance various components of the MVS pipeline. Formally, our approach involves infusing cross-view information into the pre-trained DINOv2 model to facilitate MVS learning. Furthermore, we employ different attention mechanisms for the feature encoder and cost volume regularization, focusing on feature and spatial aggregations respectively. Additionally, we uncover that some design details would substantially impact the performance of transformer modules in MVS, including normalized 3D positional encoding, adaptive attention scaling, and the position of layer normalization. Comprehensive experiments on DTU, Tanks-and-Temples, BlendedMVS, and ETH3D validate the effectiveness of the proposed method. Notably, MVSFormer++ achieves state-of-the-art performance on the challenging DTU and Tanks-and-Temples benchmarks.
This work presents Depth Anything, a highly practical solution for robust monocular depth estimation. Without pursuing novel technical modules, we aim to build a simple yet powerful foundation model dealing with any images under any circumstances. To this end, we scale up the dataset by designing a data engine to collect and automatically annotate large-scale unlabeled data (~62M), which significantly enlarges the data coverage and thus is able to reduce the generalization error. We investigate two simple yet effective strategies that make data scaling-up promising. First, a more challenging optimization target is created by leveraging data augmentation tools. It compels the model to actively seek extra visual knowledge and acquire robust representations. Second, an auxiliary supervision is developed to enforce the model to inherit rich semantic priors from pre-trained encoders. We evaluate its zero-shot capabilities extensively, including six public datasets and randomly captured photos. It demonstrates impressive generalization ability. Further, through fine-tuning it with metric depth information from NYUv2 and KITTI, new SOTAs are set. Our better depth model also results in a better depth-conditioned ControlNet. Our models are released at https://github.com/LiheYoung/Depth-Anything.
3D depth sensors using single-photon avalanche diodes (SPADs) are becoming increasingly common in applications such as autonomous navigation and object detection. Recent designs implement on-chip histogramming time-to-digital converters (TDCs) to compress the photon timestamps and reduce the bottleneck in the read-out and processing of large volumes of photon data. However, the use of full histogramming with large SPAD arrays poses significant challenges due to the associated demands in silicon area and power consumption. We propose a TDC-less dToF sensor which uses Spiking Neural Networks (SNN) to process the SPAD events directly. The proposed SNN is trained and tested on synthetic SPAD events, and while it offers five times lower precision in depth prediction than a classic centre-of-mass (CoM) algorithm (applied to histograms of the events), it achieves similar Mean Absolute Error (MAE) with faster processing speeds and significantly lower power consumption is anticipated.
An event camera is a novel vision sensor that can capture per-pixel brightness changes and output a stream of asynchronous ``events''. It has advantages over conventional cameras in those scenes with high-speed motions and challenging lighting conditions because of the high temporal resolution, high dynamic range, low bandwidth, low power consumption, and no motion blur. Therefore, several supervised monocular depth estimation from events is proposed to address scenes difficult for conventional cameras. However, depth annotation is costly and time-consuming. In this paper, to lower the annotation cost, we propose a self-supervised event-based monocular depth estimation framework named EMoDepth. EMoDepth constrains the training process using the cross-modal consistency from intensity frames that are aligned with events in the pixel coordinate. Moreover, in inference, only events are used for monocular depth prediction. Additionally, we design a multi-scale skip-connection architecture to effectively fuse features for depth estimation while maintaining high inference speed. Experiments on MVSEC and DSEC datasets demonstrate that our contributions are effective and that the accuracy can outperform existing supervised event-based and unsupervised frame-based methods.
Solutions to vision tasks in gastrointestinal endoscopy (GIE) conventionally use image encoders pretrained in a supervised manner with ImageNet-1k as backbones. However, the use of modern self-supervised pretraining algorithms and a recent dataset of 100k unlabelled GIE images (Hyperkvasir-unlabelled) may allow for improvements. In this work, we study the fine-tuned performance of models with ResNet50 and ViT-B backbones pretrained in self-supervised and supervised manners with ImageNet-1k and Hyperkvasir-unlabelled (self-supervised only) in a range of GIE vision tasks. In addition to identifying the most suitable pretraining pipeline and backbone architecture for each task, out of those considered, our results suggest three general principles. Firstly, that self-supervised pretraining generally produces more suitable backbones for GIE vision tasks than supervised pretraining. Secondly, that self-supervised pretraining with ImageNet-1k is typically more suitable than pretraining with Hyperkvasir-unlabelled, with the notable exception of monocular depth estimation in colonoscopy. Thirdly, that ViT-Bs are more suitable in polyp segmentation and monocular depth estimation in colonoscopy, ResNet50s are more suitable in polyp detection, and both architectures perform similarly in anatomical landmark recognition and pathological finding characterisation. We hope this work draws attention to the complexity of pretraining for GIE vision tasks, informs this development of more suitable approaches than the convention, and inspires further research on this topic to help advance this development. Code available: \underline{github.com/ESandML/SSL4GIE}
Purpose: Depth estimation in robotic surgery is vital in 3D reconstruction, surgical navigation and augmented reality visualization. Although the foundation model exhibits outstanding performance in many vision tasks, including depth estimation (e.g., DINOv2), recent works observed its limitations in medical and surgical domain-specific applications. This work presents a low-ranked adaptation (LoRA) of the foundation model for surgical depth estimation. Methods: We design a foundation model-based depth estimation method, referred to as Surgical-DINO, a low-rank adaptation of the DINOv2 for depth estimation in endoscopic surgery. We build LoRA layers and integrate them into DINO to adapt with surgery-specific domain knowledge instead of conventional fine-tuning. During training, we freeze the DINO image encoder, which shows excellent visual representation capacity, and only optimize the LoRA layers and depth decoder to integrate features from the surgical scene. Results: Our model is extensively validated on a MICCAI challenge dataset of SCARED, which is collected from da Vinci Xi endoscope surgery. We empirically show that Surgical-DINO significantly outperforms all the state-of-the-art models in endoscopic depth estimation tasks. The analysis with ablation studies has shown evidence of the remarkable effect of our LoRA layers and adaptation. Conclusion: Surgical-DINO shed some light on the successful adaptation of the foundation models into the surgical domain for depth estimation. There is clear evidence in the results that zero-shot prediction on pre-trained weights in computer vision datasets or naive fine-tuning is not sufficient to use the foundation model in the surgical domain directly. Code is available at https://github.com/BeileiCui/SurgicalDINO.
We introduce InseRF, a novel method for generative object insertion in the NeRF reconstructions of 3D scenes. Based on a user-provided textual description and a 2D bounding box in a reference viewpoint, InseRF generates new objects in 3D scenes. Recently, methods for 3D scene editing have been profoundly transformed, owing to the use of strong priors of text-to-image diffusion models in 3D generative modeling. Existing methods are mostly effective in editing 3D scenes via style and appearance changes or removing existing objects. Generating new objects, however, remains a challenge for such methods, which we address in this study. Specifically, we propose grounding the 3D object insertion to a 2D object insertion in a reference view of the scene. The 2D edit is then lifted to 3D using a single-view object reconstruction method. The reconstructed object is then inserted into the scene, guided by the priors of monocular depth estimation methods. We evaluate our method on various 3D scenes and provide an in-depth analysis of the proposed components. Our experiments with generative insertion of objects in several 3D scenes indicate the effectiveness of our method compared to the existing methods. InseRF is capable of controllable and 3D-consistent object insertion without requiring explicit 3D information as input. Please visit our project page at https://mohamad-shahbazi.github.io/inserf.
We present a novel approach for metric dense depth estimation based on the fusion of a single-view image and a sparse, noisy Radar point cloud. The direct fusion of heterogeneous Radar and image data, or their encodings, tends to yield dense depth maps with significant artifacts, blurred boundaries, and suboptimal accuracy. To circumvent this issue, we learn to augment versatile and robust monocular depth prediction with the dense metric scale induced from sparse and noisy Radar data. We propose a Radar-Camera framework for highly accurate and fine-detailed dense depth estimation with four stages, including monocular depth prediction, global scale alignment of monocular depth with sparse Radar points, quasi-dense scale estimation through learning the association between Radar points and image patches, and local scale refinement of dense depth using a scale map learner. Our proposed method significantly outperforms the state-of-the-art Radar-Camera depth estimation methods by reducing the mean absolute error (MAE) of depth estimation by 25.6% and 40.2% on the challenging nuScenes dataset and our self-collected ZJU-4DRadarCam dataset, respectively. Our code and dataset will be released at \url{https://github.com/MMOCKING/RadarCam-Depth}.
The capabilities of monocular depth estimation (MDE) models are limited by the availability of sufficient and diverse datasets. In the case of MDE models for autonomous driving, this issue is exacerbated by the linearity of the captured data trajectories. We propose a NeRF-based data augmentation pipeline to introduce synthetic data with more diverse viewing directions into training datasets and demonstrate the benefits of our approach to model performance and robustness. Our data augmentation pipeline, which we call \textit{NeRFmentation}, trains NeRFs on each scene in a dataset, filters out subpar NeRFs based on relevant metrics, and uses them to generate synthetic RGB-D images captured from new viewing directions. In this work, we apply our technique in conjunction with three state-of-the-art MDE architectures on the popular autonomous driving dataset, KITTI, augmenting its training set of the Eigen split. We evaluate the resulting performance gain on the original test set, a separate popular driving dataset, and our own synthetic test set.
In this paper, we introduce Hi-Map, a novel monocular dense mapping approach based on Neural Radiance Field (NeRF). Hi-Map is exceptional in its capacity to achieve efficient and high-fidelity mapping using only posed RGB inputs. Our method eliminates the need for external depth priors derived from e.g., a depth estimation model. Our key idea is to represent the scene as a hierarchical feature grid that encodes the radiance and then factorizes it into feature planes and vectors. As such, the scene representation becomes simpler and more generalizable for fast and smooth convergence on new observations. This allows for efficient computation while alleviating noise patterns by reducing the complexity of the scene representation. Buttressed by the hierarchical factorized representation, we leverage the Sign Distance Field (SDF) as a proxy of rendering for inferring the volume density, demonstrating high mapping fidelity. Moreover, we introduce a dual-path encoding strategy to strengthen the photometric cues and further boost the mapping quality, especially for the distant and textureless regions. Extensive experiments demonstrate our method's superiority in geometric and textural accuracy over the state-of-the-art NeRF-based monocular mapping methods.
We study a crucial yet often overlooked issue inherent to Vision Transformers (ViTs): feature maps of these models exhibit grid-like artifacts, which hurt the performance of ViTs in downstream dense prediction tasks such as semantic segmentation, depth prediction, and object discovery. We trace this issue down to the positional embeddings at the input stage. To mitigate this, we propose a two-stage denoising approach, termed Denoising Vision Transformers (DVT). In the first stage, we separate the clean features from those contaminated by positional artifacts by enforcing cross-view feature consistency with neural fields on a per-image basis. This per-image optimization process extracts artifact-free features from raw ViT outputs, providing clean feature estimates for offline applications. In the second stage, we train a lightweight transformer block to predict clean features from raw ViT outputs, leveraging the derived estimates of the clean features as supervision. Our method, DVT, does not require re-training the existing pre-trained ViTs, and is immediately applicable to any Vision Transformer architecture. We evaluate our method on a variety of representative ViTs (DINO, DeiT-III, EVA02, CLIP, DINOv2, DINOv2-reg) and demonstrate that DVT consistently improves existing state-of-the-art general-purpose models in semantic and geometric tasks across multiple datasets. We hope our study will encourage a re-evaluation of ViT design, especially regarding the naive use of positional embeddings. Our code and checkpoints are publicly available.