As neural radiance fields continue to advance in 3D content representation,the copyright issues surrounding 3D models oriented towards implicit representation become increasingly pressing.In response to this challenge...As neural radiance fields continue to advance in 3D content representation,the copyright issues surrounding 3D models oriented towards implicit representation become increasingly pressing.In response to this challenge,this paper treats the embedding and extraction of neural radiance field watermarks as inverse problems of image transformations and proposes a scheme for protecting neural radiance field copyrights using invertible neural network watermarking.Leveraging 2D image watermarking technology for 3D scene protection,the scheme embeds watermarks within the training images of neural radiance fields through the forward process in invertible neural networks and extracts them from images rendered by neural radiance fields through the reverse process,thereby ensuring copyright protection for both the neural radiance fields and associated 3D scenes.However,challenges such as information loss during rendering processes and deliberate tampering necessitate the design of an image quality enhancement module to increase the scheme’s robustness.This module restores distorted images through neural network processing before watermark extraction.Additionally,embedding watermarks in each training image enables watermark information extraction from multiple viewpoints.Our proposed watermarking method achieves a PSNR(Peak Signal-to-Noise Ratio)value exceeding 37 dB for images containing watermarks and 22 dB for recovered watermarked images,as evaluated on the Lego,Hotdog,and Chair datasets,respectively.These results demonstrate the efficacy of our scheme in enhancing copyright protection.展开更多
This paper presents a novel framework aimed at quantifying uncertainties associated with the 3D reconstruction of smoke from2Dimages.This approach reconstructs color and density fields from 2D images using Neural Radi...This paper presents a novel framework aimed at quantifying uncertainties associated with the 3D reconstruction of smoke from2Dimages.This approach reconstructs color and density fields from 2D images using Neural Radiance Field(NeRF)and improves image quality using frequency regularization.The NeRF model is obtained via joint training ofmultiple artificial neural networks,whereby the expectation and standard deviation of density fields and RGB values can be evaluated for each pixel.In addition,customized physics-informed neural network(PINN)with residual blocks and two-layer activation functions are utilized to input the density fields of the NeRF into Navier-Stokes equations and convection-diffusion equations to reconstruct the velocity field.The velocity uncertainties are also evaluated through ensemble learning.The effectiveness of the proposed algorithm is demonstrated through numerical examples.The presentmethod is an important step towards downstream tasks such as reliability analysis and robust optimization in engineering design.展开更多
Traditional neural radiance fields for rendering novel views require intensive input images and pre-scene optimization,which limits their practical applications.We propose a generalization method to infer scenes from ...Traditional neural radiance fields for rendering novel views require intensive input images and pre-scene optimization,which limits their practical applications.We propose a generalization method to infer scenes from input images and perform high-quality rendering without pre-scene optimization named SG-NeRF(Sparse-Input Generalized Neural Radiance Fields).Firstly,we construct an improved multi-view stereo structure based on the convolutional attention and multi-level fusion mechanism to obtain the geometric features and appearance features of the scene from the sparse input images,and then these features are aggregated by multi-head attention as the input of the neural radiance fields.This strategy of utilizing neural radiance fields to decode scene features instead of mapping positions and orientations enables our method to perform cross-scene training as well as inference,thus enabling neural radiance fields to generalize for novel view synthesis on unseen scenes.We tested the generalization ability on DTU dataset,and our PSNR(peak signal-to-noise ratio)improved by 3.14 compared with the baseline method under the same input conditions.In addition,if the scene has dense input views available,the average PSNR can be improved by 1.04 through further refinement training in a short time,and a higher quality rendering effect can be obtained.展开更多
We present SinGRAV, an attempt to learn a generative radiance volume from multi-view observations of a single natural scene, in stark contrast to existing category-level 3D generative models that learn from images of ...We present SinGRAV, an attempt to learn a generative radiance volume from multi-view observations of a single natural scene, in stark contrast to existing category-level 3D generative models that learn from images of many object-centric scenes. Inspired by SinGAN, we also learn the internal distribution of the input scene, which necessitates our key designs w.r.t. the scene representation and network architecture. Unlike popular multi-layer perceptrons (MLP)-based architectures, we particularly employ convolutional generators and discriminators, which inherently possess spatial locality bias, to operate over voxelized volumes for learning the internal distribution over a plethora of overlapping regions. On the other hand, localizing the adversarial generators and discriminators over confined areas with limited receptive fields easily leads to highly implausible geometric structures in the spatial. Our remedy is to use spatial inductive bias and joint discrimination on geometric clues in the form of 2D depth maps. This strategy is effective in improving spatial arrangement while incurring negligible additional computational cost. Experimental results 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 models, and the versatility of SinGRAV by its use in a variety of applications. Code and data will be released to facilitate further research.展开更多
Recent advancements in the field have resulted in significant progress in achieving realistic head reconstruction and manipulation using neural radiance fields(NeRF).Despite these advances,capturing intricate facial d...Recent advancements in the field have resulted in significant progress in achieving realistic head reconstruction and manipulation using neural radiance fields(NeRF).Despite these advances,capturing intricate facial details remains a persistent challenge.Moreover,casually captured input,involving both head poses and camera movements,introduces additional difficulties to existing methods of head avatar reconstruction.To address the challenge posed by video data captured with camera motion,we propose a novel method,AvatarWild,for reconstructing head avatars from monocular videos taken by consumer devices.Notably,our approach decouples the camera pose and head pose,allowing reconstructed avatars to be visualized with different poses and expressions from novel viewpoints.To enhance the visual quality of the reconstructed facial avatar,we introduce a view-dependent detail enhancement module designed to augment local facial details without compromising viewpoint consistency.Our method demonstrates superior performance compared to existing approaches,as evidenced by reconstruction and animation results on both multi-view and single-view datasets.Remarkably,our approach stands out by exclusively relying on video data captured by portable devices,such as smartphones.This not only underscores the practicality of our method but also extends its applicability to real-world scenarios where accessibility and ease of data capture are crucial.展开更多
基金supported by the National Natural Science Foundation of China,with Fund Numbers 62272478,62102451the National Defense Science and Technology Independent Research Project(Intelligent Information Hiding Technology and Its Applications in a Certain Field)and Science and Technology Innovation Team Innovative Research Project Research on Key Technologies for Intelligent Information Hiding”with Fund Number ZZKY20222102.
文摘As neural radiance fields continue to advance in 3D content representation,the copyright issues surrounding 3D models oriented towards implicit representation become increasingly pressing.In response to this challenge,this paper treats the embedding and extraction of neural radiance field watermarks as inverse problems of image transformations and proposes a scheme for protecting neural radiance field copyrights using invertible neural network watermarking.Leveraging 2D image watermarking technology for 3D scene protection,the scheme embeds watermarks within the training images of neural radiance fields through the forward process in invertible neural networks and extracts them from images rendered by neural radiance fields through the reverse process,thereby ensuring copyright protection for both the neural radiance fields and associated 3D scenes.However,challenges such as information loss during rendering processes and deliberate tampering necessitate the design of an image quality enhancement module to increase the scheme’s robustness.This module restores distorted images through neural network processing before watermark extraction.Additionally,embedding watermarks in each training image enables watermark information extraction from multiple viewpoints.Our proposed watermarking method achieves a PSNR(Peak Signal-to-Noise Ratio)value exceeding 37 dB for images containing watermarks and 22 dB for recovered watermarked images,as evaluated on the Lego,Hotdog,and Chair datasets,respectively.These results demonstrate the efficacy of our scheme in enhancing copyright protection.
基金funded by the National Natural Science Foundation of China(NSFC)(No.52274222)research project supported by Shanxi Scholarship Council of China(No.2023-036).
文摘This paper presents a novel framework aimed at quantifying uncertainties associated with the 3D reconstruction of smoke from2Dimages.This approach reconstructs color and density fields from 2D images using Neural Radiance Field(NeRF)and improves image quality using frequency regularization.The NeRF model is obtained via joint training ofmultiple artificial neural networks,whereby the expectation and standard deviation of density fields and RGB values can be evaluated for each pixel.In addition,customized physics-informed neural network(PINN)with residual blocks and two-layer activation functions are utilized to input the density fields of the NeRF into Navier-Stokes equations and convection-diffusion equations to reconstruct the velocity field.The velocity uncertainties are also evaluated through ensemble learning.The effectiveness of the proposed algorithm is demonstrated through numerical examples.The presentmethod is an important step towards downstream tasks such as reliability analysis and robust optimization in engineering design.
基金supported by the Zhengzhou Collaborative Innovation Major Project under Grant No.20XTZX06013the Henan Provincial Key Scientific Research Project of China under Grant No.22A520042。
文摘Traditional neural radiance fields for rendering novel views require intensive input images and pre-scene optimization,which limits their practical applications.We propose a generalization method to infer scenes from input images and perform high-quality rendering without pre-scene optimization named SG-NeRF(Sparse-Input Generalized Neural Radiance Fields).Firstly,we construct an improved multi-view stereo structure based on the convolutional attention and multi-level fusion mechanism to obtain the geometric features and appearance features of the scene from the sparse input images,and then these features are aggregated by multi-head attention as the input of the neural radiance fields.This strategy of utilizing neural radiance fields to decode scene features instead of mapping positions and orientations enables our method to perform cross-scene training as well as inference,thus enabling neural radiance fields to generalize for novel view synthesis on unseen scenes.We tested the generalization ability on DTU dataset,and our PSNR(peak signal-to-noise ratio)improved by 3.14 compared with the baseline method under the same input conditions.In addition,if the scene has dense input views available,the average PSNR can be improved by 1.04 through further refinement training in a short time,and a higher quality rendering effect can be obtained.
基金supported by the International(Regional)Cooperation and Exchange Program of National Natural Science Foundation of China under Grant No.62161146002the Shenzhen Collaborative Innovation Program under Grant No.CJGJZD2021048092601003.
文摘We present SinGRAV, an attempt to learn a generative radiance volume from multi-view observations of a single natural scene, in stark contrast to existing category-level 3D generative models that learn from images of many object-centric scenes. Inspired by SinGAN, we also learn the internal distribution of the input scene, which necessitates our key designs w.r.t. the scene representation and network architecture. Unlike popular multi-layer perceptrons (MLP)-based architectures, we particularly employ convolutional generators and discriminators, which inherently possess spatial locality bias, to operate over voxelized volumes for learning the internal distribution over a plethora of overlapping regions. On the other hand, localizing the adversarial generators and discriminators over confined areas with limited receptive fields easily leads to highly implausible geometric structures in the spatial. Our remedy is to use spatial inductive bias and joint discrimination on geometric clues in the form of 2D depth maps. This strategy is effective in improving spatial arrangement while incurring negligible additional computational cost. Experimental results 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 models, and the versatility of SinGRAV by its use in a variety of applications. Code and data will be released to facilitate further research.
基金supported by National Natural Science Foundation of China(No.6247075018 and No.62322210)the Innovation Funding of ICT,CAS(No.E461020)+1 种基金Beijing Munici-pal Natural Science Foundation for Distinguished Young Scholars(No.JQ21013)Beijing Municipal Science and Technology Commission(No.Z231100005923031).
文摘Recent advancements in the field have resulted in significant progress in achieving realistic head reconstruction and manipulation using neural radiance fields(NeRF).Despite these advances,capturing intricate facial details remains a persistent challenge.Moreover,casually captured input,involving both head poses and camera movements,introduces additional difficulties to existing methods of head avatar reconstruction.To address the challenge posed by video data captured with camera motion,we propose a novel method,AvatarWild,for reconstructing head avatars from monocular videos taken by consumer devices.Notably,our approach decouples the camera pose and head pose,allowing reconstructed avatars to be visualized with different poses and expressions from novel viewpoints.To enhance the visual quality of the reconstructed facial avatar,we introduce a view-dependent detail enhancement module designed to augment local facial details without compromising viewpoint consistency.Our method demonstrates superior performance compared to existing approaches,as evidenced by reconstruction and animation results on both multi-view and single-view datasets.Remarkably,our approach stands out by exclusively relying on video data captured by portable devices,such as smartphones.This not only underscores the practicality of our method but also extends its applicability to real-world scenarios where accessibility and ease of data capture are crucial.
