8,799 Hits in 3.1 sec

Neural Scene Graphs for Dynamic Scenes [article]

Julian Ost, Fahim Mannan, Nils Thuerey, Julian Knodt, Felix Heide
2021 arXiv   pre-print
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.  ...  Figure 11 : 3D Object detection with neural scene graph rendering.  ... 
arXiv:2011.10379v3 fatcat:k4lytj7clbewbjkqoao7e2js5e

AutoInt: Automatic Integration for Fast Neural Volume Rendering [article]

David B. Lindell, Julien N. P. Martel, Gordon Wetzstein
2021 arXiv   pre-print
Applying this approach to neural rendering, we improve a tradeoff between rendering speed and image quality: improving render times by greater than 10 times with a tradeoff of slightly reduced image quality  ...  Among these applications, neural volume rendering has recently been proposed as a new paradigm for view synthesis, achieving photorealistic image quality.  ...  DAGs to represent computational graphs. Our Au-toInt implementation internally maintains the computational graph of neural networks as a Directed Acyclic Graphs (DAG).  ... 
arXiv:2012.01714v2 fatcat:irl6vh3cqfhurmrspydkrv4ahy

TöRF: Time-of-Flight Radiance Fields for Dynamic Scene View Synthesis [article]

Benjamin Attal, Eliot Laidlaw, Aaron Gokaslan, Changil Kim, Christian Richardt, James Tompkin, Matthew O'Toole
2021 arXiv   pre-print
Neural networks can represent and accurately reconstruct radiance fields for static 3D scenes (e.g., NeRF).  ...  Several works extend these to dynamic scenes captured with monocular video, with promising performance.  ...  State of the art on neural rendering. Comput. Graph. Forum, 39(2):701–727, 2020.  ... 
arXiv:2109.15271v2 fatcat:bxm73ltkobaivjrxc4yv2izsoy

Semantics-Driven Remote Sensing Scene Understanding Framework for Grounded Spatio-Contextual Scene Descriptions

Abhishek V. Potnis, Surya S. Durbha, Rajat C. Shinde
2021 ISPRS International Journal of Geo-Information  
To minimize the semantic gap for remote-sensing-scene understanding, the framework puts forward the transformation of scenes by using semantic-web technologies to Remote Sensing Scene Knowledge Graphs  ...  The knowledge-graph representation of scenes has been formalized through the development of a Remote Sensing Scene Ontology(RSSO)—a core ontology for an inclusive remote-sensing-scene data product.  ...  The future directions of this research are (1) to explore neural approaches, to render natural language scene descriptions from Scene Knowledge Graphs; (2) to explore the use of inferred qualitative  ... 
doi:10.3390/ijgi10010032 fatcat:warifoopejaqfglljmkbatjrke

2D-to-3D Scene Generation and Rendering using CNN and Spatial Knowledge Representation

Gokula Nath G
2019 International Journal of Information Systems and Computer Sciences  
The new framework additionally presented a system for rendering and controlling the scene through iterative info directions.  ...  The info picture is changed over into characteristic language depiction by utilizing a neural system.  ...  The model is effective since it generates good quality image and scene rendering is also allowed by incorporating user interaction.  ... 
doi:10.30534/ijiscs/2019/25822019 fatcat:d2lu2rve5re35giat6rtio4c2i

Neural Radiance Fields for Outdoor Scene Relighting [article]

Viktor Rudnev and Mohamed Elgharib and William Smith and Lingjie Liu and Vladislav Golyanik and Christian Theobalt
2021 arXiv   pre-print
., the first approach for outdoor scene relighting based on neural radiance fields.  ...  Photorealistic editing of outdoor scenes from photographs requires a profound understanding of the image formation process and an accurate estimation of the scene geometry, reflectance and illumination  ...  At its supplied to the neural renderer as input to better capture heart is a neural radiance fields (NeRF), i.e., a neural im- scene details during relighting.  ... 
arXiv:2112.05140v1 fatcat:kgsqa4x73rcbziejajqkzrajqe

Direct Feedback Alignment Scales to Modern Deep Learning Tasks and Architectures [article]

Julien Launay, Iacopo Poli, François Boniface, Florent Krzakala
2020 arXiv   pre-print
Here, we challenge this perspective, and study the applicability of Direct Feedback Alignment to neural view synthesis, recommender systems, geometric learning, and natural language processing.  ...  Background There has been growing interest in methods capable of synthesising novel renders of a 3D scene using a dataset of past renders.  ...  We successfully learn and render real-world 3D scenes (section 3.1.1); we perform recommendation at scale (section 3.1.2); we explore graph-based citation networks (section 3.2); and we consider language  ... 
arXiv:2006.12878v2 fatcat:bisorv4p7bdebendx52ocwo6ri

Differentiable Neural Radiosity [article]

Saeed Hadadan, Matthias Zwicker
2022 arXiv   pre-print
We introduce Differentiable Neural Radiosity, a novel method of representing the solution of the differential rendering equation using a neural network.  ...  Inspired by neural radiosity techniques, we minimize the norm of the residual of the differential rendering equation to directly optimize our network.  ...  For complex rendering tasks, the graph could easily grow larger than the available GPU memory.  ... 
arXiv:2201.13190v1 fatcat:p23yv7hdqngbdiyowyil2t6m6u

Special Issue:Deep Learning in Computer Graphics

2019 IEEE Computer Graphics and Applications  
Image-based rendering synthesizes new views of a 3D scene from a set of precaptured images without any shape and material information of thescene.  ...  In the last article of this special issue, the authors represent a 3D shape as graph of small surface patches and introduce a novel graph convolutional network for 3D shape segmentation.  ... 
doi:10.1109/mcg.2019.2897385 fatcat:24mtdsnqp5d7bkraiqqvsunhr4

Neural Scene Flow Prior [article]

Xueqian Li, Jhony Kaesemodel Pontes, Simon Lucey
2021 arXiv   pre-print
A central innovation here is the inclusion of a neural scene flow prior, which uses the architecture of neural networks as a new type of implicit regularizer.  ...  Also, our neural prior's implicit and continuous scene flow representation allows us to estimate dense long-term correspondences across a sequence of point clouds.  ...  Although such deep image priors, deep mapping, and coordinate-based networks for neural scene representations have been successfully applied for inverse problems, rendering, and rigid registration, none  ... 
arXiv:2111.01253v1 fatcat:tl3afmyqjraxxilsarfghiehqe

Augmented Reality with Multilayer Occlusion [article]

Yan Feng, Yimin Chen
2006 Computer Graphics and Visual Computing  
We have designed a special scene graph tree comprised of some special nodes, namely EMO nodes.  ...  In addition, BP neural network is improved to correct the nonlinear error of magnetic sensor, consequently to detect occlusion more effectively.  ...  This node was named Background Node and was put at the uppermost and leftmost location in the scene graph tree, consequently ensured to finish rendering the real video before rendering the virtual scenes  ... 
doi:10.2312/localchapterevents/tpcg/tpcg06/099-104 dblp:conf/tpcg/FengC06 fatcat:dftnp7kyk5gbrgtvyk424xyfy4

Learning Physical Graph Representations from Visual Scenes [article]

Daniel M. Bear, Chaofei Fan, Damian Mrowca, Yunzhu Li, Seth Alter, Aran Nayebi, Jeremy Schwartz, Li Fei-Fei, Jiajun Wu, Joshua B. Tenenbaum, Daniel L.K. Yamins
2020 arXiv   pre-print
To overcome these limitations, we introduce the idea of Physical Scene Graphs (PSGs), which represent scenes as hierarchical graphs, with nodes in the hierarchy corresponding intuitively to object parts  ...  Convolutional Neural Networks (CNNs) have proved exceptional at learning representations for visual object categorization.  ...  construction, and graph rendering.  ... 
arXiv:2006.12373v2 fatcat:buvy3iaywjabfeytxrssjbwxte

Robust Point Light Source Estimation Using Differentiable Rendering [article]

Grégoire Nieto, Salma Jiddi, Philippe Robert
2018 arXiv   pre-print
Illumination estimation is often used in mixed reality to re-render a scene from another point of view, to change the color/texture of an object, or to insert a virtual object consistently lit into a real  ...  We tackle the problem of illumination retrieval given an RGBD image of the scene as an inverse problem: we aim to find the illumination that minimizes the photometric error between the rendered image and  ...  Our differentiable renderer can also be applied to improve the training of neural networks.  ... 
arXiv:1812.04857v1 fatcat:aizt5l2tmjce3od3crtgeklr4u

Structural Plan of Indoor Scenes with Personalized Preferences [article]

Xinhan Di, Pengqian Yu, Hong Zhu, Lei Cai, Qiuyan Sheng, Changyu Sun
2020 arXiv   pre-print
In particular, the model consists of the extraction of abstract graph, conditional graph generation, and conditional scene instantiation.  ...  The proposed model is able to automatically produce the layout of objects of a particular indoor scene according to property owners' preferences.  ...  The generated scenes are rendered using industrial rendering software. We remark that the rendering process is different from the previous work [10] that only produces solid color for the objects.  ... 
arXiv:2008.01323v2 fatcat:3xlw6rw5rndyzedvdxaf6qetta

DeRF: Decomposed Radiance Fields [article]

Daniel Rebain, Wei Jiang, Soroosh Yazdani, Ke Li, Kwang Moo Yi, Andrea Tagliasacchi
2020 arXiv   pre-print
With the advent of Neural Radiance Fields (NeRF), neural networks can now render novel views of a 3D scene with quality that fools the human eye.  ...  Hence, we propose to spatially decompose a scene and dedicate smaller networks for each decomposed part. When working together, these networks can render the whole scene.  ...  We render a scene (a) from a decomposed neural representation (b), consisting of a collection of spatially localized neural networks.  ... 
arXiv:2011.12490v1 fatcat:he5mkvyvrrbsfi5txa5degzo5i
« Previous Showing results 1 — 15 out of 8,799 results