No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Dynamic View Synthesis with Spatio-Temporal Feature Warping from Sparse Views
... The success of NeRF (Mildenhall et al. 2021) and subsequent works (Trevithick and Yang 2021;Wang et al. 2021b;Yu et al. 2021) have achieved impressive novel view synthesis applications. To overcome the drawback of dense input views, multiple works propose to extra regularizations or priors for sparse view novel view synthesis, such as depth and appearance smoothness (RegNeRF (Niemeyer et al. 2022), MVSNeRF ), ray entropy regularization (InfoNeRF (Kim, Seo, and Han 2022)), perceptual losses (SVS (González et al. 2022)), Spatio-Temporal consistency (Li et al. 2023) or ray distortion (Mip-NeRf360 ) et al. Besides, some recent approaches (Wei et al. 2021;Deng et al. 2022;Roessle et al. 2022) use depth priors to constrain the NeRF optimization, which also achieves promising novel view synthesis results from sparse input views. ...
The recent neural surface reconstruction approaches using volume rendering have made much progress by achieving impressive surface reconstruction quality, but are still limited to dense and highly accurate posed views. To overcome such drawbacks, this paper pays special attention on the consistent surface reconstruction from sparse views with noisy camera poses. Unlike previous approaches, the key difference of this paper is to exploit the multi-view constraints directly from the explicit geometry of the neural surface, which can be used as effective regularization to jointly learn the neural surface and refine the camera poses. To build effective multi-view constraints, we introduce a fast differentiable on-surface intersection to generate on-surface points, and propose view-consistent losses on such differentiable points to regularize the neural surface learning. Based on this point, we propose a joint learning strategy, named SC-NeuS, to perform geometry-consistent surface reconstruction in an end-to-end manner. With extensive evaluation on public datasets, our SC-NeuS can achieve consistently better surface reconstruction results with fine-grained details than previous approaches, especially from sparse and noisy camera views. The source code is available at https://github.com/zouzx/sc-neus.git.
We propose pose-guided multiplane image (MPI) synthesis which can render an animatable character in real scenes with photorealistic quality. We use a portable camera rig to capture the multi-view images along with the driving signal for the moving subject. Our method generalizes the image-to-image translation paradigm, which translates the human pose to a 3D scene representation—MPIs that can be rendered in free viewpoints, using the multi-views captures as supervision. To fully cultivate the potential of MPI, we propose depth-adaptive MPI which can be learned using variable exposure images while being robust to inaccurate camera registration. Our method demonstrates advantageous novel-view synthesis quality over the state-of-the-art approaches for characters with challenging motions. Moreover, the proposed method is generalizable to novel combinations of training poses and can be explicitly controlled. Our method achieves such expressive and animatable character rendering all in real time, serving as a promising solution for practical applications.
Neural Radiance Fields (NeRF) has demonstrated very impressive performance in novel view synthesis via implicitly modelling 3D representations from multi-view 2D images. However, most existing studies train NeRF models with either reasonable camera pose initialization or manually-crafted camera pose distributions which are often unavailable or hard to acquire in various real-world data. We design VMRF, an innovative view matching NeRF that enables effective NeRF training without requiring prior knowledge in camera poses or camera pose distributions. VMRF introduces a view matching scheme, which exploits unbalanced optimal transport to produce a feature transport plan for mapping a rendered image with randomly initialized camera pose to the corresponding real image. With the feature transport plan as the guidance, a novel pose calibration technique is designed which rectifies the initially randomized camera poses by predicting relative pose transformations between the pair of rendered and real images. Extensive experiments over a number of synthetic and real datasets show that the proposed VMRF outperforms the state-of-the-art qualitatively and quantitatively by large margins.
Ground‐based light detection and ranging (LiDAR) is becoming increasingly popular as an alternative means to conventional forest inventory methods. By gauging the distances to multiple points on the surrounding object surfaces, LiDAR acquires 3D point clouds from which tree sizes and spatial distributions can be rapidly estimated. However, the high cost and specialized skills associated with LiDAR technologies have put them out of reach for many potential users.
We here introduce ForestScanner, a free, mobile application that allows LiDAR‐based forest inventories by means of iPhone or iPad with a built‐in LiDAR sensor. ForestScanner does not require any manual analysis of 3D point clouds. As the user scans trees with an iPhone/iPad, ForestScanner estimates the stem diameters and spatial coordinates based on real‐time instance segmentation and circle fitting. The users can visualize, check and share the scanning results in situ.
By using ForestScanner, we measured the stem diameters and spatial coordinates of 672 trees within a 1 ha plot in 1 hr 39 min with an iPhone and in 1 hr 38 min with an iPad (diameter ≥ 5 cm; detection rate = 100%). The diameters measured by ForestScanner and a diameter tape were in good agreement; R ² = 0.963 for iPhone and R ² = 0.961 for iPad. ForestScanner and a conventional surveying system showed almost identical results for tree mapping (assessed by the spatial distances among trees within 0.04 ha subplots); Mantel R ² = 0.999 for both iPhone and iPad. ForestScanner reduced the person‐hours required for measuring diameters to 25.7%, mapping trees to 9.3%, and doing both to 6.8% of the person‐hours taken using a dimeter tape and the conventional surveying system.
Our results indicate that ForestScanner enables cost‐, labour‐ and time‐efficient forest inventories. The application can increase the accessibility to LiDAR for non‐experts (e.g. students, citizen scientists) and enhance resource assessments and biodiversity monitoring in forests world‐wide.
We propose a new cascaded architecture for novel view synthesis,called RGBD-Net,which consists of two core components: a hierarchical depth regression network and a depth-aware generator network. The former one predicts depth maps of the target views by using adaptive depth scaling,while the latter one leverages the predicted depths and renders spatially and temporally consistent target images. In the experimental evaluation on standard datasets,RGBD-Net not only outperforms the state-of-the-art by a clear margin,but it also generalizes well to new scenes without per-scene optimization. Moreover,we show that RGBD-Net can be optionally trained without depth supervision while still retaining high-quality rendering. Thanks to the depth regression network,RGBD-Net can be also used for creating dense 3D point clouds that are more accurate than those produced by some state-of-the-art multi-view stereo methods.
Capturing spatially-varying bidirectional reflectance distribution functions (SVBRDFs) of 3D objects with just a single, hand-held camera (such as an off-the-shelf smartphone or a DSLR camera) is a difficult, open problem. Previous works are either limited to planar geometry, or rely on previously scanned 3D geometry, thus limiting their practicality. There are several technical challenges that need to be overcome: First, the built-in flash of a camera is almost colocated with the lens, and at a fixed position; this severely hampers sampling procedures in the light-view space. Moreover, the near-field flash lights the object partially and unevenly. In terms of geometry, existing multiview stereo techniques assume diffuse reflectance only, which leads to overly smoothed 3D reconstructions, as we show in this paper. We present a simple yet powerful framework that removes the need for expensive, dedicated hardware, enabling practical acquisition of SVBRDF information from real-world, 3D objects with a single, off-the-shelf camera with a built-in flash. In addition, by removing the diffuse reflection assumption and leveraging instead such SVBRDF information, our method outputs high-quality 3D geometry reconstructions, including more accurate high-frequency details than state-of-the-art multiview stereo techniques. We formulate the joint reconstruction of SVBRDFs, shading normals, and 3D geometry as a multi-stage, iterative inverse-rendering reconstruction pipeline. Our method is also directly applicable to any existing multiview 3D reconstruction technique. We present results of captured objects with complex geometry and reflectance; we also validate our method numerically against other existing approaches that rely on dedicated hardware, additional sources of information, or both.
We present an approach to capture the 3D motion of a group of people engaged in a social interaction. The core challenges in capturing social interactions are: (1) occlusion is functional and frequent; (2) subtle motion needs to be measured over a space large enough to host a social group; (3) human appearance and configuration variation is immense; and (4) attaching markers to the body may prime the nature of interactions. The Panoptic Studio is a system organized around the thesis that social interactions should be measured through the integration of perceptual analyses over a large variety of view points. We present a modularized system designed around this principle, consisting of integrated structural, hardware, and software innovations. The system takes, as input, 480 synchronized video streams of multiple people engaged in social activities, and produces, as output, the labeled time-varying 3D structure of anatomical landmarks on individuals in the space. Our algorithm is designed to fuse the "weak" perceptual processes in the large number of views by progressively generating skeletal proposals from low-level appearance cues, and a framework for temporal refinement is also presented by associating body parts to reconstructed dense 3D trajectory stream. Our system and method are the first in reconstructing full body motion of more than five people engaged in social interactions without using markers. We also empirically demonstrate the impact of the number of views in achieving this goal.
Recently, Neural Radiance Fields (NeRF) is revolutionizing the task of novel view synthesis (NVS) for its superior performance. In this paper, we propose to synthesize dynamic scenes. Extending the methods for static scenes to dynamic scenes is not straightforward as both the scene geometry and appearance change over time, especially under monocular setup. Also, the existing dynamic NeRF methods generally require a lengthy per-scene training procedure, where multi-layer perceptrons (MLP) are fitted to model both motions and radiance. In this paper, built on top of the recent advances in voxel-grid optimization, we propose a fast deformable radiance field method to handle dynamic scenes. Our method consists of two modules. The first module adopts a deformation grid to store 3D dynamic features, and a light-weight MLP for decoding the deformation that maps a 3D point in the observation space to the canonical space using the interpolated features. The second module contains a density and a color grid to model the geometry and density of the scene. The occlusion is explicitly modeled to further improve the rendering quality. Experimental results show that our method achieves comparable performance to D-NeRF using only 20 minutes for training, which is more than 70 faster than D-NeRF, clearly demonstrating the efficiency of our proposed method.
Existing image-based rendering methods usually adopt depth-based image warping operation to synthesize novel views. In this paper, we reason the essential limitations of the traditional warping operation to be the limited neighborhood and only distance-based interpolation weights. To this end, we propose
content-aware warping
, which adaptively learns the interpolation weights for pixels of a relatively large neighborhood from their contextual information via a lightweight neural network. Based on this learnable warping module, we propose a new end-to-end learning-based framework for novel view synthesis from a set of input source views, in which two additional modules, namely confidence-based blending and feature-assistant spatial refinement, are naturally proposed to handle the occlusion issue and capture the spatial correlation among pixels of the synthesized view, respectively. Besides, we also propose a weight-smoothness loss term to regularize the network. Experimental results on light field datasets with wide baselines and multi-view datasets show that the proposed method significantly outperforms state-of-the-art methods both quantitatively and visually. The source code will be publicly available at
https://github.com/MantangGuo/CW4VS
.
Deep learning has significantly improved 2D image recognition. Extending into 3D may advance many new applications including autonomous vehicles, virtual and augmented reality, authoring 3D content, and even improving 2D recognition. However despite growing interest, 3D deep learning remains relatively underexplored. We believe that some of this disparity is due to the engineering challenges involved in 3D deep learning, such as efficiently processing heterogeneous data and reframing graphics operations to be differentiable. We address these challenges by introducing PyTorch3D, a library of modular, efficient, and differentiable operators for 3D deep learning. It includes a fast, modular differentiable renderer for meshes and point clouds, enabling analysis-by-synthesis approaches. Compared with other differentiable renderers, PyTorch3D is more modular and efficient, allowing users to more easily extend it while also gracefully scaling to large meshes and images. We compare the PyTorch3D operators and renderer with other implementations and demonstrate significant speed and memory improvements. We also use PyTorch3D to improve the state-of-the-art for unsupervised 3D mesh and point cloud prediction from 2D images on ShapeNet. PyTorch3D is open-source and we hope it will help accelerate research in 3D deep learning.
Novel view synthesis of static scenes has achieved remarkable advancements in producing photo-realistic results. However, key challenges remain for immersive rendering of dynamic scenes. One of the seminal image-based rendering method, the multi-plane image (MPI), produces high novel-view synthesis quality for static scenes. But modelling dynamic contents by MPI is not studied. In this paper, we propose a novel Temporal-MPI representation which is able to encode the rich 3D and dynamic variation information throughout the entire video as compact temporal basis and coefficients jointly learned. Time-instance MPI for rendering can be generated efficiently using mini-seconds by linear combinations of temporal basis and coefficients from Temporal-MPI. Thus novel-views at arbitrary time-instance will be able to be rendered via Temporal-MPI in real-time with high visual quality. Our method is trained and evaluated on Nvidia Dynamic Scene Dataset. We show that our proposed Temporal-MPI is much faster and more compact compared with other state-of-the-art dynamic scene modelling methods.
Novel view synthesis has recently been revolutionized by learning neural radiance fields directly from sparse observations. However, rendering images with this new paradigm is slow due to the fact that an accurate quadrature of the volume rendering equation requires a large number of samples for each ray. Previous work has mainly focused on speeding up the network evaluations that are associated with each sample point, e.g., via caching of radiance values into explicit spatial data structures, but this comes at the expense of model compactness. In this paper, we propose a novel dual-network architecture that takes an orthogonal direction by learning how to best reduce the number of required sample points. To this end, we split our network into a sampling and shading network that are jointly trained. Our training scheme employs fixed sample positions along each ray, and incrementally introduces sparsity throughout training to achieve high quality even at low sample counts. After fine-tuning with the target number of samples, the resulting compact neural representation can be rendered in real-time. Our experiments demonstrate that our approach outperforms concurrent compact neural representations in terms of quality and frame rate and performs on par with highly efficient hybrid representations. Code and supplementary material is available at https://thomasneff.github.io/adanerf.KeywordsNeural renderingNeural radiance fieldsView synthesis
Capturing and faithfully rendering photorealistic humans from novel views is a fundamental problem for AR/VR applications. While prior work has shown impressive performance capture results in laboratory settings, it is non-trivial to achieve casual free-viewpoint human capture and rendering for unseen identities with high fidelity, especially for facial expressions, hands, and clothes. To tackle these challenges we introduce a novel view synthesis framework that generates realistic renders from unseen views of any human captured from a single-view and sparse RGB-D sensor, similar to a low-cost depth camera, and without actor-specific models. We propose an architecture to create dense feature maps in novel views obtained by sphere-based neural rendering, and create complete renders using a global context inpainting model. Additionally, an enhancer network leverages the overall fidelity, even in occluded areas from the original view, producing crisp renders with fine details. We show that our method generates high-quality novel views of synthetic and real human actors given a single-stream, sparse RGB-D input. It generalizes to unseen identities, and new poses and faithfully reconstructs facial expressions. Our approach outperforms prior view synthesis methods and is robust to different levels of depth sparsity.
Structure from motion (SfM) enables us to reconstruct a scene via casual capture from cameras at different viewpoints, and novel view synthesis (NVS) allows us to render a captured scene from a new viewpoint. Both are hard with casual capture and dynamic scenes: SfM produces noisy and spatio-temporally sparse reconstructed point clouds, resulting in NVS with spatio-temporally inconsistent effects. We consider SfM and NVS parts together to ease the challenge. First, for SfM, we recover stable camera poses, then we defer the requirement for temporally-consistent points across the scene and reconstruct only a sparse point cloud per timestep that is noisy in space–time. Second, for NVS, we present a variational diffusion formulation on depths and colors that lets us robustly cope with the noise by enforcing spatio-temporal consistency via per-pixel reprojection weights derived from the input views. Together, this deferred approach lets us generate novel views for dynamic scenes without requiring challenging spatio-temporally consistent reconstructions nor training complex models on large datasets. We demonstrate our algorithm on real-world dynamic scenes against classic and more recent learning-based baseline approaches.
We present a method that achieves state-of-the-art results for synthesizing novel views of complex scenes by optimizing an underlying continuous volumetric scene function using a sparse set of input views. Our algorithm represents a scene using a fully connected (nonconvolutional) deep network, whose input is a single continuous 5D coordinate (spatial location ( x , y , z ) and viewing direction ( θ, ϕ )) and whose output is the volume density and view-dependent emitted radiance at that spatial location. We synthesize views by querying 5D coordinates along camera rays and use classic volume rendering techniques to project the output colors and densities into an image. Because volume rendering is naturally differentiable, the only input required to optimize our representation is a set of images with known camera poses. We describe how to effectively optimize neural radiance fields to render photorealistic novel views of scenes with complicated geometry and appearance, and demonstrate results that outperform prior work on neural rendering and view synthesis.
Generating free-viewpoint videos is critical for immersive VR/AR experience, but recent neural advances still lack the editing ability to manipulate the visual perception for large dynamic scenes. To fill this gap, in this paper, we propose the first approach for editable free-viewpoint video generation for large-scale view-dependent dynamic scenes using only 16 cameras. The core of our approach is a new layered neural representation, where each dynamic entity, including the environment itself, is formulated into a spatio-temporal coherent neural layered radiance representation called ST-NeRF. Such a layered representation supports manipulations of the dynamic scene while still supporting a wide free viewing experience. In our ST-NeRF, we represent the dynamic entity/layer as a continuous function, which achieves the disentanglement of location, deformation as well as the appearance of the dynamic entity in a continuous and self-supervised manner. We propose a scene parsing 4D label map tracking to disentangle the spatial information explicitly and a continuous deform module to disentangle the temporal motion implicitly. An object-aware volume rendering scheme is further introduced for the re-assembling of all the neural layers. We adopt a novel layered loss and motion-aware ray sampling strategy to enable efficient training for a large dynamic scene with multiple performers, Our framework further enables a variety of editing functions, i.e., manipulating the scale and location, duplicating or retiming individual neural layers to create numerous visual effects while preserving high realism. Extensive experiments demonstrate the effectiveness of our approach to achieve high-quality, photo-realistic, and editable free-viewpoint video generation for dynamic scenes.
We introduce Recurrent All-Pairs Field Transforms (RAFT), a new deep network architecture for optical flow. RAFT extracts per-pixel features, builds multi-scale 4D correlation volumes for all pairs of pixels, and iteratively updates a flow field through a recurrent unit that performs lookups on the correlation volumes. RAFT achieves state-of-the-art performance. On KITTI, RAFT achieves an F1-all error of 5.10%, a 16% error reduction from the best published result (6.10%). On Sintel (final pass), RAFT obtains an end-point-error of 2.855 pixels, a 30% error reduction from the best published result (4.098 pixels). In addition, RAFT has strong cross-dataset generalization as well as high efficiency in inference time, training speed, and parameter count. Code is available at https://github.com/princeton-vl/RAFT.
This paper proposes a disocclusion inpainting framework for depth-based view synthesis. It consists of four modules: foreground extraction, motion compensation, improved background reconstruction, and inpainting. The foreground extraction module detects the foreground objects and removes them from both depth map and rendered video; the motion compensation module guarantees the background reconstruction model to suit for moving camera scenarios; the improved background reconstruction module constructs a stable background video by exploiting the temporal correlation information in both 2D video and its corresponding depth map; and the constructed background video and inpainting module is used to eliminate the holes in the synthesized view. The analysis and experiment indicate that the proposed framework has good generality, scalability and effectiveness, which means most of the existing background reconstruction methods and image inpainting methods can be employed or extended as the modules in our framework. Our comparison results have demonstrated that the proposed framework achieves better synthesized quality, temporal consistency, and has lower running time compared to the other methods.
Free-viewpoint image-based rendering (IBR) is a standing challenge. IBR methods combine warped versions of input photos to synthesize a novel view. The image quality of this combination is directly affected by geometric inaccuracies of multi-view stereo (MVS) reconstruction and by view- and image-dependent effects that produce artifacts when contributions from different input views are blended. We present a new deep learning approach to blending for IBR, in which we use held-out real image data to learn blending weights to combine input photo contributions. Our Deep Blending method requires us to address several challenges to achieve our goal of interactive free-viewpoint IBR navigation. We first need to provide sufficiently accurate geometry so the Convolutional Neural Network (CNN) can succeed in finding correct blending weights. We do this by combining two different MVS reconstructions with complementary accuracy vs. completeness tradeoffs. To tightly integrate learning in an interactive IBR system, we need to adapt our rendering algorithm to produce a fixed number of input layers that can then be blended by the CNN. We generate training data with a variety of captured scenes, using each input photo as ground truth in a held-out approach. We also design the network architecture and the training loss to provide high quality novel view synthesis, while reducing temporal flickering artifacts. Our results demonstrate free-viewpoint IBR in a wide variety of scenes, clearly surpassing previous methods in visual quality, especially when moving far from the input cameras.
Learning based methods have shown very promising results for the task of depth estimation in single images. However, most existing approaches treat depth prediction as a supervised regression problem and as a result, require vast quantities of corresponding ground truth depth data for training. Just recording quality depth data in a range of environments is a challenging problem. In this paper, we innovate beyond existing approaches, replacing the use of explicit depth data during training with easier-to-obtain binocular stereo footage. We propose a novel training objective that enables our convolutional neural network to learn to perform single image depth estimation, despite the absence of ground truth depth data. By exploiting epipolar geometry constraints, we generate disparity images by training our networks with an image reconstruction loss. We show that solving for image reconstruction alone results in poor quality depth images. To overcome this problem, we propose a novel training loss that enforces consistency between the disparities produced relative to both the left and right images, leading to improved performance and robustness compared to existing approaches. Our method produces state of the art results for monocular depth estimation on the KITTI driving dataset, even outperforming supervised methods that have been trained with ground truth depth.
We present an approach to synthesizing photographic images conditioned on semantic layouts. Given a semantic label map, our approach produces an image with photographic appearance that conforms to the input layout. The approach thus functions as a rendering engine that takes a two-dimensional semantic specification of the scene and produces a corresponding photographic image. Unlike recent and contemporaneous work, our approach does not rely on adversarial training. We show that photographic images can be synthesized from semantic layouts by a single feedforward network with appropriate structure, trained end-to-end with a direct regression objective. The presented approach scales seamlessly to high resolutions; we demonstrate this by synthesizing photographic images at 2-megapixel resolution, the full resolution of our training data. Extensive perceptual experiments on datasets of outdoor and indoor scenes demonstrate that images synthesized by the presented approach are considerably more realistic than alternative approaches. The results are shown in the supplementary video at https://youtu.be/0fhUJT21-bs
Learning based methods have shown very promising results for the task of depth estimation in single images. However, most existing approaches treat depth prediction as a supervised regression problem and as a result, require vast quantities of corresponding ground truth depth data for training. Just recording quality depth data in a range of environments is a challenging problem. In this paper, we innovate beyond existing approaches, replacing the use of explicit depth data during training with easier-to-obtain binocular stereo footage. We propose a novel training objective that enables our convolutional neural network to learn to perform single image depth estimation, despite the absence of ground truth depth data. By exploiting epipolar geometry constraints, we generate disparity images by training our networks with an image reconstruction loss. We show that solving for image reconstruction alone results in poor quality depth images. To overcome this problem, we propose a novel training loss that enforces consistency between the disparities produced relative to both the left and right images, leading to improved performance and robustness compared to existing approaches. Our method produces state of the art results for monocular depth estimation on the KITTI driving dataset, even outperforming supervised methods that have been trained with ground truth depth.
There is large consent that successful training of deep networks requires many thousand annotated training samples. In this paper, we present a network and training strategy that relies on the strong use of data augmentation to use the available annotated samples more efficiently. The architecture consists of a contracting path to capture context and a symmetric expanding path that enables precise localization. We show that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks. Using the same network trained on transmitted light microscopy images (phase contrast and DIC) we won the ISBI cell tracking challenge 2015 in these categories by a large margin. Moreover, the network is fast. Segmentation of a 512x512 image takes less than a second on a recent GPU. The full implementation (based on Caffe) and the trained networks are available at
http://lmb.informatik.uni-freiburg.de/people/ronneber/u-net
.