all | dynamic | editing | fast | generalization | human | video | lighting | reconstruction | texture | semantic | pose-slam | others
- Masked Wavelet Representation for Compact Neural Radiance Fields |
[code]Neural radiance fields (NeRF) have demonstrated the potential of coordinate-based neural representation (neural fields or implicit neural representation) in neural rendering. However, using a multi-layer perceptron (MLP) to represent a 3D scene or object requires enormous computational resources and time. There have been recent studies on how to reduce these computational inefficiencies by using additional data structures, such as grids or trees. Despite the promising performance, the explicit data structure necessitates a substantial amount of memory. In this work, we present a method to reduce the size without compromising the advantages of having additional data structures. In detail, we propose using the wavelet transform on grid-based neural fields. Grid-based neural fields are for fast convergence, and the wavelet transform, whose efficiency has been demonstrated in high-performance standard codecs, is to improve the parameter efficiency of grids. Furthermore, in order to achieve a higher sparsity of grid coefficients while maintaining reconstruction quality, we present a novel trainable masking approach. Experimental results demonstrate that non-spatial grid coefficients, such as wavelet coefficients, are capable of attaining a higher level of sparsity than spatial grid coefficients, resulting in a more compact representation. With our proposed mask and compression pipeline, we achieved state-of-the-art performance within a memory budget of 2 MB. Our code is available at this https URL.
- 4K-NeRF: High Fidelity Neural Radiance Fields at Ultra High Resolutions |
[code]In this paper, we present a novel and effective framework, named 4K-NeRF, to pursue high fidelity view synthesis on the challenging scenarios of ultra high resolutions, building on the methodology of neural radiance fields (NeRF). The rendering procedure of NeRF-based methods typically relies on a pixel wise manner in which rays (or pixels) are treated independently on both training and inference phases, limiting its representational ability on describing subtle details especially when lifting to a extremely high resolution. We address the issue by better exploring ray correlation for enhancing high-frequency details benefiting from the use of geometry-aware local context. Particularly, we use the view-consistent encoder to model geometric information effectively in a lower resolution space and recover fine details through the view-consistent decoder, conditioned on ray features and depths estimated by the encoder. Joint training with patch-based sampling further facilitates our method incorporating the supervision from perception oriented regularization beyond pixel wise loss. Quantitative and qualitative comparisons with modern NeRF methods demonstrate that our method can significantly boost rendering quality for retaining high-frequency details, achieving the state-of-the-art visual quality on 4K ultra-high-resolution scenario. Code Available at \url{this https URL}
- 3D-TOGO: Towards Text-Guided Cross-Category 3D Object Generation, AAAI2023 | [code]
Text-guided 3D object generation aims to generate 3D objects described by user-defined captions, which paves a flexible way to visualize what we imagined. Although some works have been devoted to solving this challenging task, these works either utilize some explicit 3D representations (e.g., mesh), which lack texture and require post-processing for rendering photo-realistic views; or require individual time-consuming optimization for every single case. Here, we make the first attempt to achieve generic text-guided cross-category 3D object generation via a new 3D-TOGO model, which integrates a text-to-views generation module and a views-to-3D generation module. The text-to-views generation module is designed to generate different views of the target 3D object given an input caption. prior-guidance, caption-guidance and view contrastive learning are proposed for achieving better view-consistency and caption similarity. Meanwhile, a pixelNeRF model is adopted for the views-to-3D generation module to obtain the implicit 3D neural representation from the previously-generated views. Our 3D-TOGO model generates 3D objects in the form of the neural radiance field with good texture and requires no time-cost optimization for every single caption. Besides, 3D-TOGO can control the category, color and shape of generated 3D objects with the input caption. Extensive experiments on the largest 3D object dataset (i.e., ABO) are conducted to verify that 3D-TOGO can better generate high-quality 3D objects according to the input captions across 98 different categories, in terms of PSNR, SSIM, LPIPS and CLIP-score, compared with text-NeRF and Dreamfields.
- SinGRAF: Learning a 3D Generative Radiance Field for a Single Scene | [code]
Generative models have shown great promise in synthesizing photorealistic 3D objects, but they require large amounts of training data. We introduce SinGRAF, a 3D-aware generative model that is trained with a few input images of a single scene. Once trained, SinGRAF generates different realizations of this 3D scene that preserve the appearance of the input while varying scene layout. For this purpose, we build on recent progress in 3D GAN architectures and introduce a novel progressive-scale patch discrimination approach during training. With several experiments, we demonstrate that the results produced by SinGRAF outperform the closest related works in both quality and diversity by a large margin.
- NeAF: Learning Neural Angle Fields for Point Normal Estimation, AAAI2023 |
[code]Normal estimation for unstructured point clouds is an important task in 3D computer vision. Current methods achieve encouraging results by mapping local patches to normal vectors or learning local surface fitting using neural networks. However, these methods are not generalized well to unseen scenarios and are sensitive to parameter settings. To resolve these issues, we propose an implicit function to learn an angle field around the normal of each point in the spherical coordinate system, which is dubbed as Neural Angle Fields (NeAF). Instead of directly predicting the normal of an input point, we predict the angle offset between the ground truth normal and a randomly sampled query normal. This strategy pushes the network to observe more diverse samples, which leads to higher prediction accuracy in a more robust manner. To predict normals from the learned angle fields at inference time, we randomly sample query vectors in a unit spherical space and take the vectors with minimal angle values as the predicted normals. To further leverage the prior learned by NeAF, we propose to refine the predicted normal vectors by minimizing the angle offsets. The experimental results with synthetic data and real scans show significant improvements over the state-of-the-art under widely used benchmarks.
- SNAF: Sparse-view CBCT Reconstruction with Neural Attenuation Fields | [code]
Cone beam computed tomography (CBCT) has been widely used in clinical practice, especially in dental clinics, while the radiation dose of X-rays when capturing has been a long concern in CBCT imaging. Several research works have been proposed to reconstruct high-quality CBCT images from sparse-view 2D projections, but the current state-of-the-arts suffer from artifacts and the lack of fine details. In this paper, we propose SNAF for sparse-view CBCT reconstruction by learning the neural attenuation fields, where we have invented a novel view augmentation strategy to overcome the challenges introduced by insufficient data from sparse input views. Our approach achieves superior performance in terms of high reconstruction quality (30+ PSNR) with only 20 input views (25 times fewer than clinical collections), which outperforms the state-of-the-arts. We have further conducted comprehensive experiments and ablation analysis to validate the effectiveness of our approach.
- Reconstructing Hand-Held Objects from Monocular Video, SIGGRAPH-Asia2022 | [code]
This paper presents an approach that reconstructs a hand-held object from a monocular video. In contrast to many recent methods that directly predict object geometry by a trained network, the proposed approach does not require any learned prior about the object and is able to recover more accurate and detailed object geometry. The key idea is that the hand motion naturally provides multiple views of the object and the motion can be reliably estimated by a hand pose tracker. Then, the object geometry can be recovered by solving a multi-view reconstruction problem. We devise an implicit neural representation-based method to solve the reconstruction problem and address the issues of imprecise hand pose estimation, relative hand-object motion, and insufficient geometry optimization for small objects. We also provide a newly collected dataset with 3D ground truth to validate the proposed approach. The dataset and code will be released at https://dihuangdh.github.io/hhor.
- A Light Touch Approach to Teaching Transformers Multi-view Geometry | [code]
Transformers are powerful visual learners, in large part due to their conspicuous lack of manually-specified priors. This flexibility can be problematic in tasks that involve multiple-view geometry, due to the near-infinite possible variations in 3D shapes and viewpoints (requiring flexibility), and the precise nature of projective geometry (obeying rigid laws). To resolve this conundrum, we propose a "light touch" approach, guiding visual Transformers to learn multiple-view geometry but allowing them to break free when needed. We achieve this by using epipolar lines to guide the Transformer's cross-attention maps, penalizing attention values outside the epipolar lines and encouraging higher attention along these lines since they contain geometrically plausible matches. Unlike previous methods, our proposal does not require any camera pose information at test-time. We focus on pose-invariant object instance retrieval, where standard Transformer networks struggle, due to the large differences in viewpoint between query and retrieved images. Experimentally, our method outperforms state-of-the-art approaches at object retrieval, without needing pose information at test-time.
- Non-uniform Sampling Strategies for NeRF on 360° images , BMVC2022 | [code]
In recent years, the performance of novel view synthesis using perspective images has dramatically improved with the advent of neural radiance fields (NeRF). This study proposes two novel techniques that effectively build NeRF for 360{\textdegree} omnidirectional images. Due to the characteristics of a 360{\textdegree} image of ERP format that has spatial distortion in their high latitude regions and a 360{\textdegree} wide viewing angle, NeRF's general ray sampling strategy is ineffective. Hence, the view synthesis accuracy of NeRF is limited and learning is not efficient. We propose two non-uniform ray sampling schemes for NeRF to suit 360{\textdegree} images - distortion-aware ray sampling and content-aware ray sampling. We created an evaluation dataset Synth360 using Replica and SceneCity models of indoor and outdoor scenes, respectively. In experiments, we show that our proposal successfully builds 360{\textdegree} image NeRF in terms of both accuracy and efficiency. The proposal is widely applicable to advanced variants of NeRF. DietNeRF, AugNeRF, and NeRF++ combined with the proposed techniques further improve the performance. Moreover, we show that our proposed method enhances the quality of real-world scenes in 360{\textdegree} images. Synth360: this https URL.
- Unsupervised Continual Semantic Adaptation through Neural Rendering | [code]
An increasing amount of applications rely on data-driven models that are deployed for perception tasks across a sequence of scenes. Due to the mismatch between training and deployment data, adapting the model on the new scenes is often crucial to obtain good performance. In this work, we study continual multi-scene adaptation for the task of semantic segmentation, assuming that no ground-truth labels are available during deployment and that performance on the previous scenes should be maintained. We propose training a Semantic-NeRF network for each scene by fusing the predictions of a segmentation model and then using the view-consistent rendered semantic labels as pseudo-labels to adapt the model. Through joint training with the segmentation model, the Semantic-NeRF model effectively enables 2D-3D knowledge transfer. Furthermore, due to its compact size, it can be stored in a long-term memory and subsequently used to render data from arbitrary viewpoints to reduce forgetting. We evaluate our approach on ScanNet, where we outperform both a voxel-based baseline and a state-of-the-art unsupervised domain adaptation method.
- DiffusionSDF: Conditional Generative Modeling of Signed Distance Functions | [code]
Probabilistic diffusion models have achieved state-of-the-art results for image synthesis, inpainting, and text-to-image tasks. However, they are still in the early stages of generating complex 3D shapes. This work proposes DiffusionSDF, a generative model for shape completion, single-view reconstruction, and reconstruction of real-scanned point clouds. We use neural signed distance functions (SDFs) as our 3D representation to parameterize the geometry of various signals (e.g., point clouds, 2D images) through neural networks. Neural SDFs are implicit functions and diffusing them amounts to learning the reversal of their neural network weights, which we solve using a custom modulation module. Extensive experiments show that our method is capable of both realistic unconditional generation and conditional generation from partial inputs. This work expands the domain of diffusion models from learning 2D, explicit representations, to 3D, implicit representations.
- BAD-NeRF: Bundle Adjusted Deblur Neural Radiance Fields | [code]
Neural Radiance Fields (NeRF) have received considerable attention recently, due to its impressive capability in photo-realistic 3D reconstruction and novel view synthesis, given a set of posed camera images. Earlier work usually assumes the input images are in good quality. However, image degradation (e.g. image motion blur in low-light conditions) can easily happen in real-world scenarios, which would further affect the rendering quality of NeRF. In this paper, we present a novel bundle adjusted deblur Neural Radiance Fields (BAD-NeRF), which can be robust to severe motion blurred images and inaccurate camera poses. Our approach models the physical image formation process of a motion blurred image, and jointly learns the parameters of NeRF and recovers the camera motion trajectories during exposure time. In experiments, we show that by directly modeling the real physical image formation process, BAD-NeRF achieves superior performance over prior works on both synthetic and real datasets.
- OReX: Object Reconstruction from Planner Cross-sections Using Neural Fields | [code]
Reconstructing 3D shapes from planar cross-sections is a challenge inspired by downstream applications like medical imaging and geographic informatics. The input is an in/out indicator function fully defined on a sparse collection of planes in space, and the output is an interpolation of the indicator function to the entire volume. Previous works addressing this sparse and ill-posed problem either produce low quality results, or rely on additional priors such as target topology, appearance information, or input normal directions. In this paper, we present OReX, a method for 3D shape reconstruction from slices alone, featuring a Neural Field as the interpolation prior. A simple neural network is trained on the input planes to receive a 3D coordinate and return an inside/outside estimate for the query point. This prior is powerful in inducing smoothness and self-similarities. The main challenge for this approach is high-frequency details, as the neural prior is overly smoothing. To alleviate this, we offer an iterative estimation architecture and a hierarchical input sampling scheme that encourage coarse-to-fine training, allowing focusing on high frequencies at later stages. In addition, we identify and analyze a common ripple-like effect stemming from the mesh extraction step. We mitigate it by regularizing the spatial gradients of the indicator function around input in/out boundaries, cutting the problem at the root.
- Real-Time Omnidirectional Roaming in Large Scale Indoor Scenes, SIGGRAPH-Asia2022 | [code]
Neural radiance field (NeRF) has recently achieved impressive results in novel view synthesis. However, previous works on NeRF mainly focus on object-centric scenarios. They would suffer observable performance degradation in outward-facing and large-scale scenes due to limiting positional encoding capacity. To narrow the gap, we explore radiance fields in a geometry-aware fashion. We estimate explicit geometry from the omnidirectional neural radiance field that was learned from multiple 360° images. Relying on the recovered geometry, we use an adaptive divide-and-conquer strategy to slim and fine-tune the radiance fields and further improve render speed and quality. Quantitative and qualitative comparisons among baselines illustrated our predominant performance in large-scale indoor scenes and our system supports real-time VR roaming.
- AligNeRF: High-Fidelity Neural Radiance Fields via Alignment-Aware Training | [code]
Neural Radiance Fields (NeRFs) are a powerful representation for modeling a 3D scene as a continuous function. Though NeRF is able to render complex 3D scenes with view-dependent effects, few efforts have been devoted to exploring its limits in a high-resolution setting. Specifically, existing NeRF-based methods face several limitations when reconstructing high-resolution real scenes, including a very large number of parameters, misaligned input data, and overly smooth details. In this work, we conduct the first pilot study on training NeRF with high-resolution data and propose the corresponding solutions: 1) marrying the multilayer perceptron (MLP) with convolutional layers which can encode more neighborhood information while reducing the total number of parameters; 2) a novel training strategy to address misalignment caused by moving objects or small camera calibration errors; and 3) a high-frequency aware loss. Our approach is nearly free without introducing obvious training/testing costs, while experiments on different datasets demonstrate that it can recover more high-frequency details compared with the current state-of-the-art NeRF models. Project page: \url{this https URL.}
- 3DLatNav: Navigating Generative Latent Spaces for Semantic-Aware 3D Object Manipulation | [code]
3D generative models have been recently successful in generating realistic 3D objects in the form of point clouds. However, most models do not offer controllability to manipulate the shape semantics of component object parts without extensive semantic attribute labels or other reference point clouds. Moreover, beyond the ability to perform simple latent vector arithmetic or interpolations, there is a lack of understanding of how part-level semantics of 3D shapes are encoded in their corresponding generative latent spaces. In this paper, we propose 3DLatNav; a novel approach to navigating pretrained generative latent spaces to enable controlled part-level semantic manipulation of 3D objects. First, we propose a part-level weakly-supervised shape semantics identification mechanism using latent representations of 3D shapes. Then, we transfer that knowledge to a pretrained 3D object generative latent space to unravel disentangled embeddings to represent different shape semantics of component parts of an object in the form of linear subspaces, despite the unavailability of part-level labels during the training. Finally, we utilize those identified subspaces to show that controllable 3D object part manipulation can be achieved by applying the proposed framework to any pretrained 3D generative model. With two novel quantitative metrics to evaluate the consistency and localization accuracy of part-level manipulations, we show that 3DLatNav outperforms existing unsupervised latent disentanglement methods in identifying latent directions that encode part-level shape semantics of 3D objects. With multiple ablation studies and testing on state-of-the-art generative models, we show that 3DLatNav can implement controlled part-level semantic manipulations on an input point cloud while preserving other features and the realistic nature of the object.
- AsyncNeRF: Learning Large-scale Radiance Fields from Asynchronous RGB-D Sequences with Time-Pose Function | [code]
Large-scale radiance fields are promising mapping tools for smart transportation applications like autonomous driving or drone delivery. But for large-scale scenes, compact synchronized RGB-D cameras are not applicable due to limited sensing range, and using separate RGB and depth sensors inevitably leads to unsynchronized sequences. Inspired by the recent success of self-calibrating radiance field training methods that do not require known intrinsic or extrinsic parameters, we propose the first solution that self-calibrates the mismatch between RGB and depth frames. We leverage the important domain-specific fact that RGB and depth frames are actually sampled from the same trajectory and develop a novel implicit network called the time-pose function. Combining it with a large-scale radiance field leads to an architecture that cascades two implicit representation networks. To validate its effectiveness, we construct a diverse and photorealistic dataset that covers various RGB-D mismatch scenarios. Through a comprehensive benchmarking on this dataset, we demonstrate the flexibility of our method in different scenarios and superior performance over applicable prior counterparts. Codes, data, and models will be made publicly available.
- NeXT: Towards High Quality Neural Radiance Fields via Multi-skip Transformer, ECCV2022 |
[code]Neural Radiance Fields (NeRF) methods show impressive performance for novel view synthesis by representing a scene via a neural network. However, most existing NeRF based methods, including its variants, treat each sample point individually as input, while ignoring the inherent relationships between adjacent sample points from the corresponding rays, thus hindering the reconstruction performance. To address this issue, we explore a brand new scheme, namely NeXT, introducing a multi-skip transformer to capture the rich relationships between various sample points in a ray-level query. Specifically, ray tokenization is proposed to represent each ray as a sequence of point embeddings which is taken as input of our proposed NeXT. In this way, relationships between sample points are captured via the built-in self-attention mechanism to promote the reconstruction. Besides, our proposed NeXT can be easily combined with other NeRF based methods to improve their rendering quality. Extensive experiments conducted on three datasets demonstrate that NeXT significantly outperforms all previous state-of-the-art work by a large margin. In particular, the proposed NeXT surpasses the strong NeRF baseline by 2.74 dB of PSNR on Blender dataset. The code is available at https://github.com/Crishawy/NeXT.
- QRF: Implicit Neural Representations with Quantum Radiance Fields | [code]
Photorealistic rendering of real-world scenes is a tremendous challenge with a wide range of applications, including mixed reality (MR), and virtual reality (VR). Neural networks, which have long been investigated in the context of solving differential equations, have previously been introduced as implicit representations for photorealistic rendering. However, realistic rendering using classic computing is challenging because it requires time-consuming optical ray marching, and suffer computational bottlenecks due to the curse of dimensionality. In this paper, we propose Quantum Radiance Fields (QRF), which integrate the quantum circuit, quantum activation function, and quantum volume rendering for implicit scene representation. The results indicate that QRF not only exploits the advantage of quantum computing, such as high speed, fast convergence, and high parallelism, but also ensure high quality of volume rendering.
- HyperSound: Generating Implicit Neural Representations of Audio Signals with Hypernetworks | [code]
Implicit neural representations (INRs) are a rapidly growing research field, which provides alternative ways to represent multimedia signals. Recent applications of INRs include image super-resolution, compression of high-dimensional signals, or 3D rendering. However, these solutions usually focus on visual data, and adapting them to the audio domain is not trivial. Moreover, it requires a separately trained model for every data sample. To address this limitation, we propose HyperSound, a meta-learning method leveraging hypernetworks to produce INRs for audio signals unseen at training time. We show that our approach can reconstruct sound waves with quality comparable to other state-of-the-art models.
- Attention-based Neural Cellular Automata, NeurIPS2022 | [code]
Recent extensions of Cellular Automata (CA) have incorporated key ideas from modern deep learning, dramatically extending their capabilities and catalyzing a new family of Neural Cellular Automata (NCA) techniques. Inspired by Transformer-based architectures, our work presents a new class of attention-based NCAs formed using a spatially localized—yet globally organized—self-attention scheme. We introduce an instance of this class named Vision Transformer Cellular Automata (ViTCA). We present quantitative and qualitative results on denoising autoencoding across six benchmark datasets, comparing ViTCA to a U-Net, a U-Net-based CA baseline (UNetCA), and a Vision Transformer (ViT). When comparing across architectures configured to similar parameter complexity, ViTCA architectures yield superior performance across all benchmarks and for nearly every evaluation metric. We present an ablation study on various architectural configurations of ViTCA, an analysis of its effect on cell states, and an investigation on its inductive biases. Finally, we examine its learned representations via linear probes on its converged cell state hidden representations, yielding, on average, superior results when compared to our U-Net, ViT, and UNetCA baselines.
- NeX360: Real-time All-around View Synthesis with Neural Basis Expansion, TPAMI2022 | [code]
We present NeX, a new approach to novel view synthesis based on enhancements of multiplane images (MPI) that can reproduce view-dependent effects in real time. Unlike traditional MPI, our technique parameterizes each pixel as a linear combination of spherical basis functions learned from a neural network to model view-dependent effects and uses a hybrid implicit-explicit modeling strategy to improve fine detail. Moreover, we also present an extension to NeX, which leverages knowledge distillation to train multiple MPIs for unbounded 360 ∘ scenes. Our method is evaluated on several benchmark datasets: NeRF-Synthetic dataset, Light Field dataset, Real Forward-Facing dataset, Space dataset, as well as Shiny , our new dataset that contains significantly more challenging view-dependent effects, such as the rainbow reflections on the CD. Our method outperforms other real-time rendering approaches on PSNR, SSIM, and LPIPS and can render unbounded 360 ∘ scenes in real time.
- NeRFPlayer: A Streamable Dynamic Scene Representation with Decomposed Neural Radiance Fields | [code]
Visually exploring in a real-world 4D spatiotemporal space freely in VR has been a long-term quest. The task is especially appealing when only a few or even single RGB cameras are used for capturing the dynamic scene. To this end, we present an efficient framework capable of fast reconstruction, compact modeling, and streamable rendering. First, we propose to decompose the 4D spatiotemporal space according to temporal characteristics. Points in the 4D space are associated with probabilities of belonging to three categories: static, deforming, and new areas. Each area is represented and regularized by a separate neural field. Second, we propose a hybrid representations based feature streaming scheme for efficiently modeling the neural fields. Our approach, coined NeRFPlayer, is evaluated on dynamic scenes captured by single hand-held cameras and multi-camera arrays, achieving comparable or superior rendering performance in terms of quality and speed comparable to recent state-of-the-art methods, achieving reconstruction in 10 seconds per frame and real-time rendering.
- Vox-Fusion: Dense Tracking and Mapping with Voxel-based Neural Implicit Representation |
[code]In this work, we present a dense tracking and mapping system named Vox-Fusion, which seamlessly fuses neural implicit representations with traditional volumetric fusion methods. Our approach is inspired by the recently developed implicit mapping and positioning system and further extends the idea so that it can be freely applied to practical scenarios. Specifically, we leverage a voxel-based neural implicit surface representation to encode and optimize the scene inside each voxel. Furthermore, we adopt an octree-based structure to divide the scene and support dynamic expansion, enabling our system to track and map arbitrary scenes without knowing the environment like in previous works. Moreover, we proposed a high-performance multi-process framework to speed up the method, thus supporting some applications that require real-time performance. The evaluation results show that our methods can achieve better accuracy and completeness than previous methods. We also show that our Vox-Fusion can be used in augmented reality and virtual reality applications. Our source code is publicly available at this https URL.
- Compressing multidimensional weather and climate data into neural networks | [code]
Weather and climate simulations produce petabytes of high-resolution data that are later analyzed by researchers in order to understand climate change or severe weather. We propose a new method of compressing this multidimensional weather and climate data: a coordinate-based neural network is trained to overfit the data, and the resulting parameters are taken as a compact representation of the original grid-based data. While compression ratios range from 300x to more than 3,000x, our method outperforms the state-of-the-art compressor SZ3 in terms of weighted RMSE, MAE. It can faithfully preserve important large scale atmosphere structures and does not introduce artifacts. When using the resulting neural network as a 790x compressed dataloader to train the WeatherBench forecasting model, its RMSE increases by less than 2%. The three orders of magnitude compression democratizes access to high-resolution climate data and enables numerous new research directions.
- Neural Sound Field Decomposition with Super-resolution of Sound Direction | [code]
Sound field decomposition predicts waveforms in arbitrary directions using signals from a limited number of microphones as inputs. Sound field decomposition is fundamental to downstream tasks, including source localization, source separation, and spatial audio reproduction. Conventional sound field decomposition methods such as Ambisonics have limited spatial decomposition resolution. This paper proposes a learning-based Neural Sound field Decomposition (NeSD) framework to allow sound field decomposition with fine spatial direction resolution, using recordings from microphone capsules of a few microphones at arbitrary positions. The inputs of a NeSD system include microphone signals, microphone positions, and queried directions. The outputs of a NeSD include the waveform and the presence probability of a queried position. We model the NeSD systems respectively with different neural networks, including fully connected, time delay, and recurrent neural networks. We show that the NeSD systems outperform conventional Ambisonics and DOANet methods in sound field decomposition and source localization on speech, music, and sound events datasets. Demos are available at this https URL.
- Continuous conditional video synthesis by neural processes |
[code]We propose a unified model for multiple conditional video synthesis tasks, including video prediction and video frame interpolation. We show that conditional video synthesis can be formulated as a neural process, which maps input spatio-temporal coordinates to target pixel values given context spatio-temporal coordinates and pixels values. Specifically, we feed an implicit neural representations of coordinates into a Transformer-based non-autoregressive conditional video synthesis model. Our task-specific models outperform previous work for video interpolation on multiple datasets and reach a competitive performance with the state-of-the-art models for video prediction. Importantly, the model is able to interpolate or predict with an arbitrary high frame rate, i.e., continuous synthesis. Our source code is available at this https URL.
- Geometric Warping Error Aware CNN for DIBR Oriented View Synthesis, ACMMM2022 | [code]
Depth Image based Rendering (DIBR) oriented view synthesis is an important virtual view generation technique. It warps the reference view images to the target viewpoint based on their depth maps, without requiring many available viewpoints. However, in the 3D warping process, pixels are warped to fractional pixel locations and then rounded (or interpolated) to integer pixels, resulting in geometric warping error and reducing the image quality. This resembles, to some extent, the image super-resolution problem, but with unfixed fractional pixel locations. To address this problem, we propose a geometric warping error aware CNN (GWEA) framework to enhance the DIBR oriented view synthesis. First, a deformable convolution based geometric warping error aware alignment (GWEA-DCA) module is developed, by taking advantage of the geometric warping error preserved in the DIBR module. The offset learned in the deformable convolution can account for the geometric warping error to facilitate the mapping from the fractional pixels to integer pixels. Moreover, in view that the pixels in the warped images are of different qualities due to the different strengths of warping errors, an attention enhanced view blending (GWEA-AttVB) module is further developed to adaptively fuse the pixels from different warped images. Finally, a partial convolution based hole filling and refinement module fills the remaining holes and improves the quality of the overall image. Experiments show that our model can synthesize higher-quality images than the existing methods, and ablation study is also conducted, validating the effectiveness of each proposed module.
- ViewFool: Evaluating the Robustness of Visual Recognition to Adversarial Viewpoints, NeurIPS2022 | [code]
Recent studies have demonstrated that visual recognition models lack robustness to distribution shift. However, current work mainly considers model robustness to 2D image transformations, leaving viewpoint changes in the 3D world less explored. In general, viewpoint changes are prevalent in various real-world applications (e.g., autonomous driving), making it imperative to evaluate viewpoint robustness. In this paper, we propose a novel method called ViewFool to find adversarial viewpoints that mislead visual recognition models. By encoding real-world objects as neural radiance fields (NeRF), ViewFool characterizes a distribution of diverse adversarial viewpoints under an entropic regularizer, which helps to handle the fluctuations of the real camera pose and mitigate the reality gap between the real objects and their neural representations. Experiments validate that the common image classifiers are extremely vulnerable to the generated adversarial viewpoints, which also exhibit high cross-model transferability. Based on ViewFool, we introduce ImageNet-V, a new out-of-distribution dataset for benchmarking viewpoint robustness of image classifiers. Evaluation results on 40 classifiers with diverse architectures, objective functions, and data augmentations reveal a significant drop in model performance when tested on ImageNet-V, which provides a possibility to leverage ViewFool as an effective data augmentation strategy to improve viewpoint robustness.
- Novel View Synthesis for Surgical Recording | [code]
Recording surgery in operating rooms is one of the essential tasks for education and evaluation of medical treatment. However, recording the fields which depict the surgery is difficult because the targets are heavily occluded during surgery by the heads or hands of doctors or nurses. We use a recording system which multiple cameras embedded in the surgical lamp, assuming that at least one camera is recording the target without occlusion. In this paper, we propose Conditional-BARF (C-BARF) to generate occlusion-free images by synthesizing novel view images from the camera, aiming to generate videos with smooth camera pose transitions. To the best of our knowledge, this is the first work to tackle the problem of synthesizing a novel view image from multiple images for the surgery scene. We conduct experiments using an original dataset of three different types of surgeries. Our experiments show that we can successfully synthesize novel views from the images recorded by the multiple cameras embedded in the surgical lamp.
- Dfferentiable Raycasting for Self-supervised Occupancy Forecasting, ECCV2022 |
[code]Motion planning for safe autonomous driving requires learning how the environment around an ego-vehicle evolves with time. Ego-centric perception of driveable regions in a scene not only changes with the motion of actors in the environment, but also with the movement of the ego-vehicle itself. Self-supervised representations proposed for large-scale planning, such as ego-centric freespace, confound these two motions, making the representation difficult to use for downstream motion planners. In this paper, we use geometric occupancy as a natural alternative to view-dependent representations such as freespace. Occupancy maps naturally disentangle the motion of the environment from the motion of the ego-vehicle. However, one cannot directly observe the full 3D occupancy of a scene (due to occlusion), making it difficult to use as a signal for learning. Our key insight is to use differentiable raycasting to "render" future occupancy predictions into future LiDAR sweep predictions, which can be compared with ground-truth sweeps for self-supervised learning. The use of differentiable raycasting allows occupancy to emerge as an internal representation within the forecasting network. In the absence of groundtruth occupancy, we quantitatively evaluate the forecasting of raycasted LiDAR sweeps and show improvements of upto 15 F1 points. For downstream motion planners, where emergent occupancy can be directly used to guide non-driveable regions, this representation relatively reduces the number of collisions with objects by up to 17% as compared to freespace-centric motion planners.
- Self-improving Multiplane-to-layer Images for Novel View Synthesis, WACV2023 |
[code]We present a new method for lightweight novel-view synthesis that generalizes to an arbitrary forward-facing scene. Recent approaches are computationally expensive, require per-scene optimization, or produce a memory-expensive representation. We start by representing the scene with a set of fronto-parallel semitransparent planes and afterward convert them to deformable layers in an end-to-end manner. Additionally, we employ a feed-forward refinement procedure that corrects the estimated representation by aggregating information from input views. Our method does not require fine-tuning when a new scene is processed and can handle an arbitrary number of views without restrictions. Experimental results show that our approach surpasses recent models in terms of common metrics and human evaluation, with the noticeable advantage in inference speed and compactness of the inferred layered geometry, see this https URL
- NARF22: Neural Articulated Radiance Fields for Configuration-Aware Rendering, IROS2022 | [code]
Articulated objects pose a unique challenge for robotic perception and manipulation. Their increased number of degrees-of-freedom makes tasks such as localization computationally difficult, while also making the process of real-world dataset collection unscalable. With the aim of addressing these scalability issues, we propose Neural Articulated Radiance Fields (NARF22), a pipeline which uses a fully-differentiable, configuration-parameterized Neural Radiance Field (NeRF) as a means of providing high quality renderings of articulated objects. NARF22 requires no explicit knowledge of the object structure at inference time. We propose a two-stage parts-based training mechanism which allows the object rendering models to generalize well across the configuration space even if the underlying training data has as few as one configuration represented. We demonstrate the efficacy of NARF22 by training configurable renderers on a real-world articulated tool dataset collected via a Fetch mobile manipulation robot. We show the applicability of the model to gradient-based inference methods through a configuration estimation and 6 degree-of-freedom pose refinement task. The project webpage is available at: this https URL.
- SinGRAV: Learning a Generative Radiance Volume from a Single Natural Scene | [code]
We present a 3D generative model for general natural scenes. Lacking necessary volumes of 3D data characterizing the target scene, we propose to learn from a single scene. Our key insight is that a natural scene often contains multiple constituents whose geometry, texture, and spatial arrangements follow some clear patterns, but still exhibit rich variations over different regions within the same scene. This suggests localizing the learning of a generative model on substantial local regions. Hence, we exploit a multi-scale convolutional network, which possesses the spatial locality bias in nature, to learn from the statistics of local regions at multiple scales within a single scene. In contrast to existing methods, our learning setup bypasses the need to collect data from many homogeneous 3D scenes for learning common features. We coin our method SinGRAV, for learning a Generative RAdiance Volume from a Single natural scene. We demonstrate the ability of SinGRAV in generating plausible and diverse variations from a single scene, the merits of SinGRAV over state-of-the-art generative neural scene methods, as well as the versatility of SinGRAV by its use in a variety of applications, spanning 3D scene editing, composition, and animation. Code and data will be released to facilitate further research.
- IntrinsicNeRF: Learning Intrinsic Neural Radiance Fields for Editable Novel View Synthesis |
[code]We present intrinsic neural radiance fields, dubbed IntrinsicNeRF, that introduce intrinsic decomposition into the NeRF-based~\cite{mildenhall2020nerf} neural rendering method and can perform editable novel view synthesis in room-scale scenes while existing inverse rendering combined with neural rendering methods~\cite{zhang2021physg, zhang2022modeling} can only work on object-specific scenes. Given that intrinsic decomposition is a fundamentally ambiguous and under-constrained inverse problem, we propose a novel distance-aware point sampling and adaptive reflectance iterative clustering optimization method that enables IntrinsicNeRF with traditional intrinsic decomposition constraints to be trained in an unsupervised manner, resulting in temporally consistent intrinsic decomposition results. To cope with the problem of different adjacent instances of similar reflectance in a scene being incorrectly clustered together, we further propose a hierarchical clustering method with coarse-to-fine optimization to obtain a fast hierarchical indexing representation. It enables compelling real-time augmented reality applications such as scene recoloring, material editing, and illumination variation. Extensive experiments on Blender Object and Replica Scene demonstrate that we can obtain high-quality, consistent intrinsic decomposition results and high-fidelity novel view synthesis even for challenging sequences. Code and data are available on the project webpage: this https URL.
- SCI: A spectrum concentrated implicit neural compression for biomedical data | [code]
Massive collection and explosive growth of the huge amount of medical data, demands effective compression for efficient storage, transmission and sharing. Readily available visual data compression techniques have been studied extensively but tailored for nature images/videos, and thus show limited performance on medical data which are of different characteristics. Emerging implicit neural representation (INR) is gaining momentum and demonstrates high promise for fitting diverse visual data in target-data-specific manner, but a general compression scheme covering diverse medical data is so far absent. To address this issue, we firstly derive a mathematical explanation for INR's spectrum concentration property and an analytical insight on the design of compression-oriented INR architecture. Further, we design a funnel shaped neural network capable of covering broad spectrum of complex medical data and achieving high compression ratio. Based on this design, we conduct compression via optimization under given budget and propose an adaptive compression approach SCI, which adaptively partitions the target data into blocks matching the concentrated spectrum envelop of the adopted INR, and allocates parameter with high representation accuracy under given compression ratio. The experiments show SCI's superior performance over conventional techniques and wide applicability across diverse medical data.
- Distilling Style from Image Pairs for Global Forward and Inverse Tone Mapping, CVMP2022 | [code]
Many image enhancement or editing operations, such as forward and inverse tone mapping or color grading, do not have a unique solution, but instead a range of solutions, each representing a different style. Despite this, existing learning-based methods attempt to learn a unique mapping, disregarding this style. In this work, we show that information about the style can be distilled from collections of image pairs and encoded into a 2- or 3-dimensional vector. This gives us not only an efficient representation but also an interpretable latent space for editing the image style. We represent the global color mapping between a pair of images as a custom normalizing flow, conditioned on a polynomial basis of the pixel color. We show that such a network is more effective than PCA or VAE at encoding image style in low-dimensional space and lets us obtain an accuracy close to 40 dB, which is about 7-10 dB improvement over the state-of-the-art methods.
- Towards Multi-spatiotemporal-scale Generalized PDE Modeling | [code]
Partial differential equations (PDEs) are central to describing complex physical system simulations. Their expensive solution techniques have led to an increased interest in deep neural network based surrogates. However, the practical utility of training such surrogates is contingent on their ability to model complex multi-scale spatio-temporal phenomena. Various neural network architectures have been proposed to target such phenomena, most notably Fourier Neural Operators (FNOs) which give a natural handle over local & global spatial information via parameterization of different Fourier modes, and U-Nets which treat local and global information via downsampling and upsampling paths. However, generalizing across different equation parameters or different time-scales still remains a challenge. In this work, we make a comprehensive comparison between various FNO and U-Net like approaches on fluid mechanics problems in both vorticity-stream and velocity function form. For U-Nets, we transfer recent architectural improvements from computer vision, most notably from object segmentation and generative modeling. We further analyze the design considerations for using FNO layers to improve performance of U-Net architectures without major degradation of computational performance. Finally, we show promising results on generalization to different PDE parameters and time-scales with a single surrogate model.
- Implicit Neural Spatial Representations for Time-dependent PDEs | [code]
Numerically solving partial differential equations (PDEs) often entails spatial and temporal discretizations. Traditional methods (e.g., finite difference, finite element, smoothed-particle hydrodynamics) frequently adopt explicit spatial discretizations, such as grids, meshes, and point clouds, where each degree-of-freedom corresponds to a location in space. While these explicit spatial correspondences are intuitive to model and understand, these representations are not necessarily optimal for accuracy, memory-usage, or adaptivity. In this work, we explore implicit neural representation as an alternative spatial discretization, where spatial information is implicitly stored in the neural network weights. With implicit neural spatial representation, PDE-constrained time-stepping translates into updating neural network weights, which naturally integrates with commonly adopted optimization time integrators. We validate our approach on a variety of classic PDEs with examples involving large elastic deformations, turbulent fluids, and multiscale phenomena. While slower to compute than traditional representations, our approach exhibits higher accuracy, lower memory consumption, and dynamically adaptive allocation of degrees of freedom without complex remeshing.
- Continuous PDE Dynamics Forecasting with Implicit Neural Representations | [code]
Effective data-driven PDE forecasting methods often rely on fixed spatial and / or temporal discretizations. This raises limitations in real-world applications like weather prediction where flexible extrapolation at arbitrary spatiotemporal locations is required. We address this problem by introducing a new data-driven approach, DINo, that models a PDE's flow with continuous-time dynamics of spatially continuous functions. This is achieved by embedding spatial observations independently of their discretization via Implicit Neural Representations in a small latent space temporally driven by a learned ODE. This separate and flexible treatment of time and space makes DINo the first data-driven model to combine the following advantages. It extrapolates at arbitrary spatial and temporal locations; it can learn from sparse irregular grids or manifolds; at test time, it generalizes to new grids or resolutions. DINo outperforms alternative neural PDE forecasters in a variety of challenging generalization scenarios on representative PDE systems.
- Towards General-Purpose Representation Learning of Polygonal Geometries, GeoInformatica | [code]
Neural network representation learning for spatial data is a common need for geographic artificial intelligence (GeoAI) problems. In recent years, many advancements have been made in representation learning for points, polylines, and networks, whereas little progress has been made for polygons, especially complex polygonal geometries. In this work, we focus on developing a general-purpose polygon encoding model, which can encode a polygonal geometry (with or without holes, single or multipolygons) into an embedding space. The result embeddings can be leveraged directly (or finetuned) for downstream tasks such as shape classification, spatial relation prediction, and so on. To achieve model generalizability guarantees, we identify a few desirable properties: loop origin invariance, trivial vertex invariance, part permutation invariance, and topology awareness. We explore two different designs for the encoder: one derives all representations in the spatial domain; the other leverages spectral domain representations. For the spatial domain approach, we propose ResNet1D, a 1D CNN-based polygon encoder, which uses circular padding to achieve loop origin invariance on simple polygons. For the spectral domain approach, we develop NUFTspec based on Non-Uniform Fourier Transformation (NUFT), which naturally satisfies all the desired properties. We conduct experiments on two tasks: 1) shape classification based on MNIST; 2) spatial relation prediction based on two new datasets - DBSR-46K and DBSR-cplx46K. Our results show that NUFTspec and ResNet1D outperform multiple existing baselines with significant margins. While ResNet1D suffers from model performance degradation after shape-invariance geometry modifications, NUFTspec is very robust to these modifications due to the nature of the NUFT.
- Enforcing safety for vision-based controllers via Control Barrier Functions and Neural Radiance Fields | [code]
To navigate complex environments, robots must increasingly use high-dimensional visual feedback (e.g. images) for control. However, relying on high-dimensional image data to make control decisions raises important questions; particularly, how might we prove the safety of a visual-feedback controller? Control barrier functions (CBFs) are powerful tools for certifying the safety of feedback controllers in the state-feedback setting, but CBFs have traditionally been poorly-suited to visual feedback control due to the need to predict future observations in order to evaluate the barrier function. In this work, we solve this issue by leveraging recent advances in neural radiance fields (NeRFs), which learn implicit representations of 3D scenes and can render images from previously-unseen camera perspectives, to provide single-step visual foresight for a CBF-based controller. This novel combination is able to filter out unsafe actions and intervene to preserve safety. We demonstrate the effect of our controller in real-time simulation experiments where it successfully prevents the robot from taking dangerous actions.
- WaterNeRF: Neural Radiance Fields for Underwater Scenes | [code]
Underwater imaging is a critical task performed by marine robots for a wide range of applications including aquaculture, marine infrastructure inspection, and environmental monitoring. However, water column effects, such as attenuation and backscattering, drastically change the color and quality of imagery captured underwater. Due to varying water conditions and range-dependency of these effects, restoring underwater imagery is a challenging problem. This impacts downstream perception tasks including depth estimation and 3D reconstruction. In this paper, we advance state-of-the-art in neural radiance fields (NeRFs) to enable physics-informed dense depth estimation and color correction. Our proposed method, WaterNeRF, estimates parameters of a physics-based model for underwater image formation, leading to a hybrid data-driven and model-based solution. After determining the scene structure and radiance field, we can produce novel views of degraded as well as corrected underwater images, along with dense depth of the scene. We evaluate the proposed method qualitatively and quantitatively on a real underwater dataset.
- How Does It Feel? Self-Supervised Costmap Learning for Off-Road Vehicle Traversability | [code]
Estimating terrain traversability in off-road environments requires reasoning about complex interaction dynamics between the robot and these terrains. However, it is challenging to build an accurate physics model, or create informative labels to learn a model in a supervised manner, for these interactions. We propose a method that learns to predict traversability costmaps by combining exteroceptive environmental information with proprioceptive terrain interaction feedback in a self-supervised manner. Additionally, we propose a novel way of incorporating robot velocity in the costmap prediction pipeline. We validate our method in multiple short and large-scale navigation tasks on a large, autonomous all-terrain vehicle (ATV) on challenging off-road terrains, and demonstrate ease of integration on a separate large ground robot. Our short-scale navigation results show that using our learned costmaps leads to overall smoother navigation, and provides the robot with a more fine-grained understanding of the interactions between the robot and different terrain types, such as grass and gravel. Our large-scale navigation trials show that we can reduce the number of interventions by up to 57% compared to an occupancy-based navigation baseline in challenging off-road courses ranging from 400 m to 3150 m.
- wildNeRF: Complete view synthesis of in-the-wild dynamic scenes captured using sparse monocular data | [code]
We present a novel neural radiance model that is trainable in a self-supervised manner for novel-view synthesis of dynamic unstructured scenes. Our end-to-end trainable algorithm learns highly complex, real-world static scenes within seconds and dynamic scenes with both rigid and non-rigid motion within minutes. By differentiating between static and motion-centric pixels, we create high-quality representations from a sparse set of images. We perform extensive qualitative and quantitative evaluation on existing benchmarks and set the state-of-the-art on performance measures on the challenging NVIDIA Dynamic Scenes Dataset. Additionally, we evaluate our model performance on challenging real-world datasets such as Cholec80 and SurgicalActions160.
- Density-aware NeRF Ensembles: Quantifying Predictive Uncertainty in Neural Radiance Fields | [code]
We show that ensembling effectively quantifies model uncertainty in Neural Radiance Fields (NeRFs) if a density-aware epistemic uncertainty term is considered. The naive ensembles investigated in prior work simply average rendered RGB images to quantify the model uncertainty caused by conflicting explanations of the observed scene. In contrast, we additionally consider the termination probabilities along individual rays to identify epistemic model uncertainty due to a lack of knowledge about the parts of a scene unobserved during training. We achieve new state-of-the-art performance across established uncertainty quantification benchmarks for NeRFs, outperforming methods that require complex changes to the NeRF architecture and training regime. We furthermore demonstrate that NeRF uncertainty can be utilised for next-best view selection and model refinement.
- LATITUDE: Robotic Global Localization with Truncated Dynamic Low-pass Filter in City-scale NeRF, ICRA2023 |
[code]Neural Radiance Fields (NeRFs) have made great success in representing complex 3D scenes with high-resolution details and efficient memory. Nevertheless, current NeRF-based pose estimators have no initial pose prediction and are prone to local optima during optimization. In this paper, we present LATITUDE: Global Localization with Truncated Dynamic Low-pass Filter, which introduces a two-stage localization mechanism in city-scale NeRF. In place recognition stage, we train a regressor through images generated from trained NeRFs, which provides an initial value for global localization. In pose optimization stage, we minimize the residual between the observed image and rendered image by directly optimizing the pose on tangent plane. To avoid convergence to local optimum, we introduce a Truncated Dynamic Low-pass Filter (TDLF) for coarse-to-fine pose registration. We evaluate our method on both synthetic and real-world data and show its potential applications for high-precision navigation in large-scale city scenes. Codes and data will be publicly available at this https URL.
- Implicit Neural Representations for Medical Imaging Segmentation, MICCAI2022 | [code]
3D signals in medical imaging, such as CT scans, are usually parameterized as a discrete grid of voxels. For instance, existing state-of-the-art organ segmentation methods learn discrete segmentation maps. Unfortunately, the memory requirements of such methods grow cubically with increasing spatial resolution, which makes them unsuitable for processing high resolution scans. To overcome this, we design an Implicit Organ Segmentation Network (IOSNet) that utilizes continuous Implicit Neural Representations and has several useful properties. Firstly, the IOSNet decoder memory is roughly constant and independent of the spatial resolution since it parameterizes the segmentation map as a continuous function. Secondly, IOSNet converges much faster than discrete voxel based methods due to its ability to accurately segment organs irrespective of organ sizes, thereby alleviating size imbalance issues without requiring any auxiliary tricks. Thirdly, IOSNet naturally supports super-resolution (i.e. sampling at arbitrary resolutions during inference) due to its continuous learnt representations. Moreover, despite using a simple lightweight decoder, IOSNet consistently outperforms the discrete specialized segmentation architecture UNet. Hence, our approach demonstrates that Implicit Neural Representations are well-suited for medical imaging applications, especially for processing high-resolution 3D medical scans.
- DevNet: Self-supervised Monocular Depth Learning via Density Volume Construction, ECCV2022 | [code]
Self-supervised depth learning from monocular images normally relies on the 2D pixel-wise photometric relation between temporally adjacent image frames. However, they neither fully exploit the 3D point-wise geometric correspondences, nor effectively tackle the ambiguities in the photometric warping caused by occlusions or illumination inconsistency. To address these problems, this work proposes Density Volume Construction Network (DevNet), a novel self-supervised monocular depth learning framework, that can consider 3D spatial information, and exploit stronger geometric constraints among adjacent camera frustums. Instead of directly regressing the pixel value from a single image, our DevNet divides the camera frustum into multiple parallel planes and predicts the pointwise occlusion probability density on each plane. The final depth map is generated by integrating the density along corresponding rays. During the training process, novel regularization strategies and loss functions are introduced to mitigate photometric ambiguities and overfitting. Without obviously enlarging model parameters size or running time, DevNet outperforms several representative baselines on both the KITTI-2015 outdoor dataset and NYU-V2 indoor dataset. In particular, the root-mean-square-deviation is reduced by around 4% with DevNet on both KITTI-2015 and NYU-V2 in the task of depth estimation. Code is available at this https URL.
- Learning A Unified 3D Point Cloud for View Synthesis | [code]
3D point cloud representation-based view synthesis methods have demonstrated effectiveness. However, existing methods usually synthesize novel views only from a single source view, and it is non-trivial to generalize them to handle multiple source views for pursuing higher reconstruction quality. In this paper, we propose a new deep learning-based view synthesis paradigm, which learns a unified 3D point cloud from different source views. Specifically, we first construct sub-point clouds by projecting source views to 3D space based on their depth maps. Then, we learn the unified 3D point cloud by adaptively fusing points at a local neighborhood defined on the union of the sub-point clouds. Besides, we also propose a 3D geometry-guided image restoration module to fill the holes and recover high-frequency details of the rendered novel views. Experimental results on three benchmark datasets demonstrate that our method outperforms state-of-the-art view synthesis methods to a large extent both quantitatively and visually.
- Self-Supervised Coordinate Projection Network for Sparse-View Computed Tomography | [code]
In the present work, we propose a Self-supervised COordinate Projection nEtwork (SCOPE) to reconstruct the artifacts-free CT image from a single SV sinogram by solving the inverse tomography imaging problem. Compared with recent related works that solve similar problems using implicit neural representation network (INR), our essential contribution is an effective and simple re-projection strategy that pushes the tomography image reconstruction quality over supervised deep learning CT reconstruction works. The proposed strategy is inspired by the simple relationship between linear algebra and inverse problems. To solve the under-determined linear equation system, we first introduce INR to constrain the solution space via image continuity prior and achieve a rough solution. And secondly, we propose to generate a dense view sinogram that improves the rank of the linear equation system and produces a more stable CT image solution space. Our experiment results demonstrate that the re-projection strategy significantly improves the image reconstruction quality (+3 dB for PSNR at least). Besides, we integrate the recent hash encoding into our SCOPE model, which greatly accelerates the model training. Finally, we evaluate SCOPE in parallel and fan X-ray beam SVCT reconstruction tasks. Experimental results indicate that the proposed SCOPE model outperforms two latest INR-based methods and two well-popular supervised DL methods quantitatively and qualitatively.
- CU-Net: Efficient Point Cloud Color Upsampling Network | [code]
Point cloud upsampling is necessary for Augmented Reality, Virtual Reality, and telepresence scenarios. Although the geometry upsampling is well studied to densify point cloud coordinates, the upsampling of colors has been largely overlooked. In this paper, we propose CU-Net, the first deep-learning point cloud color upsampling model. Leveraging a feature extractor based on sparse convolution and a color prediction module based on neural implicit function, CU-Net achieves linear time and space complexity. Therefore, CU-Net is theoretically guaranteed to be more efficient than most existing methods with quadratic complexity. Experimental results demonstrate that CU-Net can colorize a photo-realistic point cloud with nearly a million points in real time, while having better visual quality than baselines. Besides, CU-Net can adapt to an arbitrary upsampling ratio and unseen objects. Our source code will be released to the public soon.
- Implicit Full Waveform Inversion with Deep Neural Representation | [code]
Full waveform inversion (FWI) commonly stands for the state-of-the-art approach for imaging subsurface structures and physical parameters, however, its implementation usually faces great challenges, such as building a good initial model to escape from local minima, and evaluating the uncertainty of inversion results. In this paper, we propose the implicit full waveform inversion (IFWI) algorithm using continuously and implicitly defined deep neural representations. Compared to FWI, which is sensitive to the initial model, IFWI benefits from the increased degrees of freedom with deep learning optimization, thus allowing to start from a random initialization, which greatly reduces the risk of non-uniqueness and being trapped in local minima. Both theoretical and experimental analyses indicates that, given a random initial model, IFWI is able to converge to the global minimum and produce a high-resolution image of subsurface with fine structures. In addition, uncertainty analysis of IFWI can be easily performed by approximating Bayesian inference with various deep learning approaches, which is analyzed in this paper by adding dropout neurons. Furthermore, IFWI has a certain degree of robustness and strong generalization ability that are exemplified in the experiments of various 2D geological models. With proper setup, IFWI can also be well suited for multi-scale joint geophysical inversion.
- FoV-NeRF: Foveated Neural Radiance Fields for Virtual Reality, TVCG2022 | [code]
Virtual Reality (VR) is becoming ubiquitous with the rise of consumer displays and commercial VR platforms. Such displays require low latency and high quality rendering of synthetic imagery with reduced compute overheads. Recent advances in neural rendering showed promise of unlocking new possibilities in 3D computer graphics via image-based representations of virtual or physical environments. Specifically, the neural radiance fields (NeRF) demonstrated that photo-realistic quality and continuous view changes of 3D scenes can be achieved without loss of view-dependent effects. While NeRF can significantly benefit rendering for VR applications, it faces unique challenges posed by high field-of-view, high resolution, and stereoscopic/egocentric viewing, typically causing low quality and high latency of the rendered images. In VR, this not only harms the interaction experience but may also cause sickness. To tackle these problems toward six-degrees-of-freedom, egocentric, and stereo NeRF in VR, we present the first gaze-contingent 3D neural representation and view synthesis method . We incorporate the human psychophysics of visual- and stereo-acuity into an egocentric neural representation of 3D scenery. We then jointly optimize the latency/performance and visual quality while mutually bridging human perception and neural scene synthesis to achieve perceptually high-quality immersive interaction. We conducted both objective analysis and subjective studies to evaluate the effectiveness of our approach. We find that our method significantly reduces latency (up to 99% time reduction compared with NeRF) without loss of high-fidelity rendering (perceptually identical to full-resolution ground truth). The presented approach may serve as the first step toward future VR/AR systems that capture, teleport, and visualize remote environments in real-time.
- CLONeR: Camera-Lidar Fusion for Occupancy Grid-aided Neural Representations | [code]
This paper proposes CLONeR, which significantly improves upon NeRF by allowing it to model large outdoor driving scenes that are observed from sparse input sensor views. This is achieved by decoupling occupancy and color learning within the NeRF framework into separate Multi-Layer Perceptrons (MLPs) trained using LiDAR and camera data, respectively. In addition, this paper proposes a novel method to build differentiable 3D Occupancy Grid Maps (OGM) alongside the NeRF model, and leverage this occupancy grid for improved sampling of points along a ray for volumetric rendering in metric space.
- HyperTime: Implicit Neural Representation for Time Series | [code]
Implicit neural representations (INRs) have recently emerged as a powerful tool that provides an accurate and resolution-independent encoding of data. Their robustness as general approximators has been shown in a wide variety of data sources, with applications on image, sound, and 3D scene representation. However, little attention has been given to leveraging these architectures for the representation and analysis of time series data. In this paper, we analyze the representation of time series using INRs, comparing different activation functions in terms of reconstruction accuracy and training convergence speed. We show how these networks can be leveraged for the imputation of time series, with applications on both univariate and multivariate data. Finally, we propose a hypernetwork architecture that leverages INRs to learn a compressed latent representation of an entire time series dataset. We introduce an FFT-based loss to guide training so that all frequencies are preserved in the time series. We show that this network can be used to encode time series as INRs, and their embeddings can be interpolated to generate new time series from existing ones. We evaluate our generative method by using it for data augmentation, and show that it is competitive against current state-of-the-art approaches for augmentation of time series.
- NIDN: Neural Inverse Design of Nanostructures | [code]
In the recent decade, computational tools have become central in material design, allowing rapid development cycles at reduced costs. Machine learning tools are especially on the rise in photonics. However, the inversion of the Maxwell equations needed for the design is particularly challenging from an optimization standpoint, requiring sophisticated software. We present an innovative, open-source software tool called Neural Inverse Design of Nanostructures (NIDN) that allows designing complex, stacked material nanostructures using a physics-based deep learning approach. Instead of a derivative-free or data-driven optimization or learning method, we perform a gradient-based neural network training where we directly optimize the material and its structure based on its spectral characteristics. NIDN supports two different solvers, rigorous coupled-wave analysis and a finite-difference time-domain method. The utility and validity of NIDN are demonstrated on several synthetic examples as well as the design of a 1550 nm filter and anti-reflection coating. Results match experimental baselines, other simulation tools, and the desired spectral characteristics. Given its full modularity in regard to network architectures and Maxwell solvers as well as open-source, permissive availability, NIDN will be able to support computational material design processes in a broad range of applications.
- Monte Carlo Denoising Using Implicit Neural Representation | [code]
Monte Carlo path tracing is a popular 3D rendering technique in computer graphics, but it often requires a costly tradeoff between the amount of noise in the image and computation time. Therefore, it is useful to attempt to “smooth out” a noisy image, typically by constructing new data between the samples or applying filters to the image. In this work, we investigate the feasibility of training a neural network to implicitly represent the radiance of a fixed-viewpoint scene as a continuous function. We implement the neural network using a multilayer perceptron network and train it on a sparsely sampled image that is generated by an offline Monte Carlo renderer. This training data uses the (x, y) coordinate of each sample on the image plane as inputs and the RGB color of the sample as outputs. Additionally, we provide the network with the surface normal, depth, and albedo of the first ray intersection as extra inputs alongside the pixel coordinates. These extra input dimensions improve the quality of the implicit representation by helping the network account for changes in depth, normal, and diffuse color. Once the network is trained on the sparsely sampled scene, we can densely sample the network many times per pixel to create the final denoised image. We find that this network can quickly learn and denoise images in scenes with soft lighting and glossy reflections, and it can easily handle discontinuities in depth, normal, and diffuse color with just a small amount of training.
- DoF-NeRF: Depth-of-Field Meets Neural Radiance Fields, ACMMM2022 |
[code]Neural Radiance Field (NeRF) and its variants have exhibited great success on representing 3D scenes and synthesizing photo-realistic novel views. However, they are generally based on the pinhole camera model and assume all-in-focus inputs. This limits their applicability as images captured from the real world often have finite depth-of-field (DoF). To mitigate this issue, we introduce DoF-NeRF, a novel neural rendering approach that can deal with shallow DoF inputs and can simulate DoF effect. In particular, it extends NeRF to simulate the aperture of lens following the principles of geometric optics. Such a physical guarantee allows DoF-NeRF to operate views with different focus configurations. Benefiting from explicit aperture modeling, DoF-NeRF also enables direct manipulation of DoF effect by adjusting virtual aperture and focus parameters. It is plug-and-play and can be inserted into NeRF-based frameworks. Experiments on synthetic and real-world datasets show that, DoF-NeRF not only performs comparably with NeRF in the all-in-focus setting, but also can synthesize all-in-focus novel views conditioned on shallow DoF inputs. An interesting application of DoF-NeRF to DoF rendering is also demonstrated.
- Neural Density-Distance Fields, ECCV2022 |
[code]The success of neural fields for 3D vision tasks is now indisputable. Following this trend, several methods aiming for visual localization (e.g., SLAM) have been proposed to estimate distance or density fields using neural fields. However, it is difficult to achieve high localization performance by only density fields-based methods such as Neural Radiance Field (NeRF) since they do not provide density gradient in most empty regions. On the other hand, distance field-based methods such as Neural Implicit Surface (NeuS) have limitations in objects' surface shapes. This paper proposes Neural Density-Distance Field (NeDDF), a novel 3D representation that reciprocally constrains the distance and density fields. We extend distance field formulation to shapes with no explicit boundary surface, such as fur or smoke, which enable explicit conversion from distance field to density field. Consistent distance and density fields realized by explicit conversion enable both robustness to initial values and high-quality registration. Furthermore, the consistency between fields allows fast convergence from sparse point clouds. Experiments show that NeDDF can achieve high localization performance while providing comparable results to NeRF on novel view synthesis. The code is available at this https URL.
- End-to-end View Synthesis via NeRF Attention | [code]
In this paper, we present a simple seq2seq formulation for view synthesis where we take a set of ray points as input and output colors corresponding to the rays. Directly applying a standard transformer on this seq2seq formulation has two limitations. First, the standard attention cannot successfully fit the volumetric rendering procedure, and therefore high-frequency components are missing in the synthesized views. Second, applying global attention to all rays and pixels is extremely inefficient. Inspired by the neural radiance field (NeRF), we propose the NeRF attention (NeRFA) to address the above problems. On the one hand, NeRFA considers the volumetric rendering equation as a soft feature modulation procedure. In this way, the feature modulation enhances the transformers with the NeRF-like inductive bias. On the other hand, NeRFA performs multi-stage attention to reduce the computational overhead. Furthermore, the NeRFA model adopts the ray and pixel transformers to learn the interactions between rays and pixels. NeRFA demonstrates superior performance over NeRF and NerFormer on four datasets: DeepVoxels, Blender, LLFF, and CO3D. Besides, NeRFA establishes a new state-of-the-art under two settings: the single-scene view synthesis and the category-centric novel view synthesis. The code will be made publicly available.
- Neural Strands: Learning Hair Geometry and Appearance from Multi-View Images, ECCV2022 |
[code]We present Neural Strands, a novel learning framework for modeling accurate hair geometry and appearance from multi-view image inputs. The learned hair model can be rendered in real-time from any viewpoint with high-fidelity view-dependent effects. Our model achieves intuitive shape and style control unlike volumetric counterparts. To enable these properties, we propose a novel hair representation based on a neural scalp texture that encodes the geometry and appearance of individual strands at each texel location. Furthermore, we introduce a novel neural rendering framework based on rasterization of the learned hair strands. Our neural rendering is strand-accurate and anti-aliased, making the rendering view-consistent and photorealistic. Combining appearance with a multi-view geometric prior, we enable, for the first time, the joint learning of appearance and explicit hair geometry from a multi-view setup. We demonstrate the efficacy of our approach in terms of fidelity and efficiency for various hairstyles.
- Neural Green’s function for Laplacian systems, Computer & Graphics | [code]
Solving linear system of equations stemming from Laplacian operators is at the heart of a wide range of applications. Due to the sparsity of the linear systems, iterative solvers such as Conjugate Gradient and Multigrid are usually employed when the solution has a large number of degrees of freedom. These iterative solvers can be seen as sparse approximations of the Green’s function for the Laplacian operator. In this paper we propose a machine learning approach that regresses a Green’s function from boundary conditions. This is enabled by a Green’s function that can be effectively represented in a multi-scale fashion, drastically reducing the cost associated with a dense matrix representation. Additionally, since the Green’s function is solely dependent on boundary conditions, training the proposed neural network does not require sampling the right-hand side of the linear system. We show results that our method outperforms state of the art Conjugate Gradient and Multigrid methods.
- On the Learnability of Physical Concepts: Can a Neural Network Understand What's Real? | [code]
We revisit the classic signal-to-symbol barrier in light of the remarkable ability of deep neural networks to generate realistic synthetic data. DeepFakes and spoofing highlight the feebleness of the link between physical reality and its abstract representation, whether learned by a digital computer or a biological agent. Starting from a widely applicable definition of abstract concept, we show that standard feed-forward architectures cannot capture but trivial concepts, regardless of the number of weights and the amount of training data, despite being extremely effective classifiers. On the other hand, architectures that incorporate recursion can represent a significantly larger class of concepts, but may still be unable to learn them from a finite dataset. We qualitatively describe the class of concepts that can be "understood" by modern architectures trained with variants of stochastic gradient descent, using a (free energy) Lagrangian to measure information complexity. Even if a concept has been understood, however, a network has no means of communicating its understanding to an external agent, except through continuous interaction and validation. We then characterize physical objects as abstract concepts and use the previous analysis to show that physical objects can be encoded by finite architectures. However, to understand physical concepts, sensors must provide persistently exciting observations, for which the ability to control the data acquisition process is essential (active perception). The importance of control depends on the modality, benefiting visual more than acoustic or chemical perception. Finally, we conclude that binding physical entities to digital identities is possible in finite time with finite resources, solving in principle the signal-to-symbol barrier problem, but we highlight the need for continuous validation.
- Plenoxels: Radiance Fields without Neural Networks, CVPR2022(oral) |
[code]We introduce Plenoxels (plenoptic voxels), a system for photorealistic view synthesis. Plenoxels represent a scene as a sparse 3D grid with spherical harmonics. This representation can be optimized from calibrated images via gradient methods and regularization without any neural components. On standard, benchmark tasks, Plenoxels are optimized two orders of magnitude faster than Neural Radiance Fields with no loss in visual quality.
- Urban Radiance Fields, CVPR2022 | [code]
The goal of this work is to perform 3D reconstruction and novel view synthesis from data captured by scanning platforms commonly deployed for world mapping in urban outdoor environments (e.g., Street View). Given a sequence of posed RGB images and lidar sweeps acquired by cameras and scanners moving through an outdoor scene, we produce a model from which 3D surfaces can be extracted and novel RGB images can be synthesized. Our approach extends Neural Radiance Fields, which has been demonstrated to synthesize realistic novel images for small scenes in controlled settings, with new methods for leveraging asynchronously captured lidar data, for addressing exposure variation between captured images, and for leveraging predicted image segmentations to supervise densities on rays pointing at the sky. Each of these three extensions provides significant performance improvements in experiments on Street View data. Our system produces state-of-the-art 3D surface reconstructions and synthesizes higher quality novel views in comparison to both traditional methods (e.g.~COLMAP) and recent neural representations (e.g.~Mip-NeRF).
- NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis, ECCV2020 |
[code]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 (non-convolutional) 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. View synthesis results are best viewed as videos, so we urge readers to view our supplementary video for convincing comparisons.
- NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collections, CVPR2021 | [code]
We present a learning-based method for synthesizing novel views of complex scenes using only unstructured collections of in-the-wild photographs. We build on Neural Radiance Fields (NeRF), which uses the weights of a multilayer perceptron to model the density and color of a scene as a function of 3D coordinates. While NeRF works well on images of static subjects captured under controlled settings, it is incapable of modeling many ubiquitous, real-world phenomena in uncontrolled images, such as variable illumination or transient occluders. We introduce a series of extensions to NeRF to address these issues, thereby enabling accurate reconstructions from unstructured image collections taken from the internet. We apply our system, dubbed NeRF-W, to internet photo collections of famous landmarks, and demonstrate temporally consistent novel view renderings that are significantly closer to photorealism than the prior state of the art.
- Ha-NeRF: Hallucinated Neural Radiance Fields in the Wild, CVPR2022 |
[code]Neural Radiance Fields (NeRF) has recently gained popularity for its impressive novel view synthesis ability. This paper studies the problem of hallucinated NeRF: i.e., recovering a realistic NeRF at a different time of day from a group of tourism images. Existing solutions adopt NeRF with a controllable appearance embedding to render novel views under various conditions, but they cannot render view-consistent images with an unseen appearance. To solve this problem, we present an end-to-end framework for constructing a hallucinated NeRF, dubbed as Ha-NeRF. Specifically, we propose an appearance hallucination module to handle time-varying appearances and transfer them to novel views. Considering the complex occlusions of tourism images, we introduce an anti-occlusion module to decompose the static subjects for visibility accurately. Experimental results on synthetic data and real tourism photo collections demonstrate that our method can hallucinate the desired appearances and render occlusion-free images from different views.
- Nerfies: Deformable Neural Radiance Fields, ICCV2021 | [code]
We present the first method capable of photorealistically reconstructing deformable scenes using photos/videos captured casually from mobile phones. Our approach augments neural radiance fields (NeRF) by optimizing an additional continuous volumetric deformation field that warps each observed point into a canonical 5D NeRF. We observe that these NeRF-like deformation fields are prone to local minima, and propose a coarse-to-fine optimization method for coordinate-based models that allows for more robust optimization. By adapting principles from geometry processing and physical simulation to NeRF-like models, we propose an elastic regularization of the deformation field that further improves robustness. We show that our method can turn casually captured selfie photos/videos into deformable NeRF models that allow for photorealistic renderings of the subject from arbitrary viewpoints, which we dub "nerfies." We evaluate our method by collecting time-synchronized data using a rig with two mobile phones, yielding train/validation images of the same pose at different viewpoints. We show that our method faithfully reconstructs non-rigidly deforming scenes and reproduces unseen views with high fidelity.
- D-NeRF: Neural Radiance Fields for Dynamic Scenes, CVPR2021 |
[code]Neural rendering techniques combining machine learning with geometric reasoning have arisen as one of the most promising approaches for synthesizing novel views of a scene from a sparse set of images. Among these, stands out the Neural radiance fields (NeRF), which trains a deep network to map 5D input coordinates (representing spatial location and viewing direction) into a volume density and view-dependent emitted radiance. However, despite achieving an unprecedented level of photorealism on the generated images, NeRF is only applicable to static scenes, where the same spatial location can be queried from different images. In this paper we introduce D-NeRF, a method that extends neural radiance fields to a dynamic domain, allowing to reconstruct and render novel images of objects under rigid and non-rigid motions from a \emph{single} camera moving around the scene. For this purpose we consider time as an additional input to the system, and split the learning process in two main stages: one that encodes the scene into a canonical space and another that maps this canonical representation into the deformed scene at a particular time. Both mappings are simultaneously learned using fully-connected networks. Once the networks are trained, D-NeRF can render novel images, controlling both the camera view and the time variable, and thus, the object movement. We demonstrate the effectiveness of our approach on scenes with objects under rigid, articulated and non-rigid motions. Code, model weights and the dynamic scenes dataset will be released.
- Dynamic Neural Radiance Fields for Monocular 4D Facial Avatar Reconstruction, CVPR2021 |
[code]We present dynamic neural radiance fields for modeling the appearance and dynamics of a human face. Digitally modeling and reconstructing a talking human is a key building-block for a variety of applications. Especially, for telepresence applications in AR or VR, a faithful reproduction of the appearance including novel viewpoint or head-poses is required. In contrast to state-of-the-art approaches that model the geometry and material properties explicitly, or are purely image-based, we introduce an implicit representation of the head based on scene representation networks. To handle the dynamics of the face, we combine our scene representation network with a low-dimensional morphable model which provides explicit control over pose and expressions. We use volumetric rendering to generate images from this hybrid representation and demonstrate that such a dynamic neural scene representation can be learned from monocular input data only, without the need of a specialized capture setup. In our experiments, we show that this learned volumetric representation allows for photo-realistic image generation that surpasses the quality of state-of-the-art video-based reenactment methods.
- PVA: Pixel-aligned Volumetric Avatars, CVPR2021 | [code]
Acquisition and rendering of photorealistic human heads is a highly challenging research problem of particular importance for virtual telepresence. Currently, the highest quality is achieved by volumetric approaches trained in a person-specific manner on multi-view data. These models better represent fine structure, such as hair, compared to simpler mesh-based models. Volumetric models typically employ a global code to represent facial expressions, such that they can be driven by a small set of animation parameters. While such architectures achieve impressive rendering quality, they can not easily be extended to the multi-identity setting. In this paper, we devise a novel approach for predicting volumetric avatars of the human head given just a small number of inputs. We enable generalization across identities by a novel parameterization that combines neural radiance fields with local, pixel-aligned features extracted directly from the inputs, thus side-stepping the need for very deep or complex networks. Our approach is trained in an end-to-end manner solely based on a photometric rerendering loss without requiring explicit 3D supervision.We demonstrate that our approach outperforms the existing state of the art in terms of quality and is able to generate faithful facial expressions in a multi-identity setting.
- Animatable Neural Radiance Fields for Human Body Modeling, ICCV2021 |
[code]This paper addresses the challenge of reconstructing an animatable human model from a multi-view video. Some recent works have proposed to decompose a non-rigidly deforming scene into a canonical neural radiance field and a set of deformation fields that map observation-space points to the canonical space, thereby enabling them to learn the dynamic scene from images. However, they represent the deformation field as translational vector field or SE(3) field, which makes the optimization highly under-constrained. Moreover, these representations cannot be explicitly controlled by input motions. Instead, we introduce neural blend weight fields to produce the deformation fields. Based on the skeleton-driven deformation, blend weight fields are used with 3D human skeletons to generate observation-to-canonical and canonical-to-observation correspondences. Since 3D human skeletons are more observable, they can regularize the learning of deformation fields. Moreover, the learned blend weight fields can be combined with input skeletal motions to generate new deformation fields to animate the human model. Experiments show that our approach significantly outperforms recent human synthesis methods. The code will be available at https://zju3dv.github.io/animatable_nerf/.
- NeRF++: Analyzing and Improving Neural Radiance Fields |
[code]Neural Radiance Fields (NeRF) achieve impressive view synthesis results for a variety of capture settings, including 360 capture of bounded scenes and forward-facing capture of bounded and unbounded scenes. NeRF fits multi-layer perceptrons (MLPs) representing view-invariant opacity and view-dependent color volumes to a set of training images, and samples novel views based on volume rendering techniques. In this technical report, we first remark on radiance fields and their potential ambiguities, namely the shape-radiance ambiguity, and analyze NeRF's success in avoiding such ambiguities. Second, we address a parametrization issue involved in applying NeRF to 360 captures of objects within large-scale, unbounded 3D scenes. Our method improves view synthesis fidelity in this challenging scenario. Code is available at this https URL.
- Neural Scene Graphs for Dynamic Scenes, CVPR2021(oral) |
[code]Recent implicit neural rendering methods have demonstrated that it is possible to learn accurate view synthesis for complex scenes by predicting their volumetric density and color supervised solely by a set of RGB images. However, existing methods are restricted to learning efficient representations of static scenes that encode all scene objects into a single neural network, and lack the ability to represent dynamic scenes and decompositions into individual scene objects. In this work, we present the first neural rendering method that decomposes dynamic scenes into scene graphs. We propose a learned scene graph representation, which encodes object transformation and radiance, to efficiently render novel arrangements and views of the scene. To this end, we learn implicitly encoded scenes, combined with a jointly learned latent representation to describe objects with a single implicit function. We assess the proposed method on synthetic and real automotive data, validating that our approach learns dynamic scenes -- only by observing a video of this scene -- and allows for rendering novel photo-realistic views of novel scene compositions with unseen sets of objects at unseen poses.
- In-Place Scene Labelling and Understanding with Implicit Scene Representation, ICCV2021(oral) |
[code]Semantic labelling is highly correlated with geometry and radiance reconstruction, as scene entities with similar shape and appearance are more likely to come from similar classes. Recent implicit neural reconstruction techniques are appealing as they do not require prior training data, but the same fully self-supervised approach is not possible for semantics because labels are human-defined properties.