Filters








41 Hits in 8.7 sec

DISN: Deep Implicit Surface Network for High-quality Single-view 3D Reconstruction [article]

Qiangeng Xu, Weiyue Wang, Duygu Ceylan, Radomir Mech, Ulrich Neumann
2021 arXiv   pre-print
In this paper, we present DISN, a Deep Implicit Surface Network which can generate a high-quality detail-rich 3D mesh from an 2D image by predicting the underlying signed distance fields.  ...  Reconstructing 3D shapes from single-view images has been a long-standing research problem.  ...  DISN: Deep Implicit Surface Network for High-quality Single-view 3D Reconstruction  ... 
arXiv:1905.10711v4 fatcat:zpgpibd3jjhynaxjahqvxgkenq

D^2IM-Net: Learning Detail Disentangled Implicit Fields from Single Images [article]

Manyi Li, Hao Zhang
2020 arXiv   pre-print
We present the first single-view 3D reconstruction network aimed at recovering geometric details from an input image which encompass both topological shape structures and surface features.  ...  Our key idea is to train the network to learn a detail disentangled reconstruction consisting of two functions, one implicit field representing the coarse 3D shape and the other capturing the details.  ...  Recently, rapid advances in deep learning have propelled the development of data-driven single-view 3D reconstruction methods.  ... 
arXiv:2012.06650v2 fatcat:2aoj55etsbgd3bscz6aq7e2s3e

Deep Optimized Priors for 3D Shape Modeling and Reconstruction [article]

Mingyue Yang, Yuxin Wen, Weikai Chen, Yongwei Chen, Kui Jia
2020 arXiv   pre-print
We introduce a new learning framework for 3D modeling and reconstruction that greatly improves the generalization ability of a deep generator.  ...  We show that the proposed strategy effectively breaks the barriers constrained by the pre-trained priors and could lead to high-quality adaptation to unseen data.  ...  Specifically, while we are able to faithfully reconstruct 3D surface with sparse or even a single view, we can also achieve similar quality of reconstruction with the traditional stereo-based approach  ... 
arXiv:2012.07241v1 fatcat:3wf6o5th3rgrjno55azlzt6uyq

Neural Implicit 3D Shapes from Single Images with Spatial Patterns [article]

Yixin Zhuang and Yunzhe Liu and Yujie Wang and Baoquan Chen
2022 arXiv   pre-print
Neural implicit functions have achieved impressive results for reconstructing 3D shapes from single images.  ...  However, the image features for describing 3D point samplings of implicit functions are less effective when significant variations of occlusions, views, and appearances exist from the image.  ...  Related Work Deep Neural Networks for SVR. There has been a lot of research on single image reconstruction task.  ... 
arXiv:2106.03087v3 fatcat:bngdpraxgfdqrldg26zaacwuz4

Ladybird: Quasi-Monte Carlo Sampling for Deep Implicit Field Based 3D Reconstruction with Symmetry [article]

Yifan Xu, Tianqi Fan, Yi Yuan, Gurprit Singh
2020 arXiv   pre-print
Deep implicit field regression methods are effective for 3D reconstruction from single-view images.  ...  Our proposed system Ladybird is able to create high quality 3D object reconstructions from a single input image.  ...  Our work advocates that Grid+FPS is suitable for 3D reconstruction based on deep implicit fields and marching cube.  ... 
arXiv:2007.13393v1 fatcat:bo2iat7ov5cdnbcdwenbkziiji

3PSDF: Three-Pole Signed Distance Function for Learning Surfaces with Arbitrary Topologies [article]

Weikai Chen, Cheng Lin, Weiyang Li, Bo Yang
2022 arXiv   pre-print
Recent advances in learning 3D shapes using neural implicit functions have achieved impressive results by breaking the previous barrier of resolution and diversity for varying topologies.  ...  Nonetheless, as their direct outputs are point clouds, robustly obtaining high-quality meshing results from discrete points remains an open question.  ...  Single-view Reconstruction on MGN We evaluate and compare the representation capability of 3PSDF, DISN [43] and OccNet [26] on MGN dataset [4] for single-view 3D reconstruction.  ... 
arXiv:2205.15572v1 fatcat:iayqem6fbrhajnb5xshc4c5acu

SkeletonNet: A Topology-Preserving Solution for Learning Mesh Reconstruction of Object Surfaces from RGB Images [article]

Jiapeng Tang, Xiaoguang Han, Mingkui Tan, Xin Tong, Kui Jia
2021 arXiv   pre-print
With the learned skeletal volumes, we propose two models, the Skeleton-Based GraphConvolutional Neural Network (SkeGCNN) and the Skeleton-Regularized Deep Implicit Surface Network (SkeDISN), which respectivelybuild  ...  This paper focuses on the challenging task of learning 3D object surface reconstructions from RGB images.  ...  shown that, by leveraging the great modeling capacities of deep networks, the 3D surface shapes of generic objects can be learned and reconstructed from as few as a single image [9] , [10] , [11] ,  ... 
arXiv:2008.05742v3 fatcat:2ha7dtqmkzc3bphf3vsskarpjq

MeshSDF: Differentiable Iso-Surface Extraction [article]

Edoardo Remelli, Artem Lukoianov, Stephan R. Richter, Benoît Guillard, Timur Bagautdinov, Pierre Baque, Pascal Fua
2020 arXiv   pre-print
We use two different applications to validate our theoretical insight: Single-View Reconstruction via Differentiable Rendering and Physically-Driven Shape Optimization.  ...  deep implicit field.  ...  This representation has been successfully used for single view reconstruction [31, 8, 60] and 3D shape completion [10] .  ... 
arXiv:2006.03997v2 fatcat:msmzxi66xrbjtjym3lylz7zlmm

SingleSketch2Mesh : Generating 3D Mesh model from Sketch [article]

Nitish Bhardwaj, Dhornala Bharadwaj, Alpana Dubey
2022 arXiv   pre-print
We propose a novel AI based ensemble approach, SingleSketch2Mesh, for generating 3D models from hand-drawn sketches.  ...  Our approach is based on Generative Networks and Encoder-Decoder Architecture to generate 3D mesh model from a hand-drawn sketch. We evaluate our solution with existing solutions.  ...  A smooth surface extraction algorithm is implemented post implicit fields to yield a high quality 3D mesh model for various use cases.  ... 
arXiv:2203.03157v3 fatcat:3etvsxyhwnh4xdmczumabm3c3y

UCLID-Net: Single View Reconstruction in Object Space [article]

Benoit Guillard, Edoardo Remelli, Pascal Fua
2020 arXiv   pre-print
Most state-of-the-art deep geometric learning single-view reconstruction approaches rely on encoder-decoder architectures that output either shape parametrizations or implicit representations.  ...  Furthermore, the single-view pipeline naturally extends to multi-view reconstruction, which we also show.  ...  Finally, the single-view pipeline naturally extends to multi-view reconstruction, which we also provide an example for.  ... 
arXiv:2006.03817v2 fatcat:yxetaoobn5cixh6wctu5xxqhmu

FvOR: Robust Joint Shape and Pose Optimization for Few-view Object Reconstruction [article]

Zhenpei Yang, Zhile Ren, Miguel Angel Bautista, Zaiwei Zhang, Qi Shan, Qixing Huang
2022 arXiv   pre-print
The core of our approach is a fast and robust multi-view reconstruction algorithm to jointly refine 3D geometry and camera pose estimation using learnable neural network modules.  ...  Reconstructing an accurate 3D object model from a few image observations remains a challenging problem in computer vision.  ...  OccNet [37] is a top performing method for single-view 3D reconstruction .  ... 
arXiv:2205.07763v1 fatcat:4cnzu2pwafdbvckrx5jeqrckzi

Geo-PIFu: Geometry and Pixel Aligned Implicit Functions for Single-view Human Reconstruction [article]

Tong He, John Collomosse, Hailin Jin, Stefano Soatto
2020 arXiv   pre-print
Our method is based on a deep implicit function-based representation to learn latent voxel features using a structure-aware 3D U-Net, to constrain the model in two ways: first, to resolve feature ambiguities  ...  We show that, by both encoding query points and constraining global shape using latent voxel features, the reconstruction we obtain for clothed human meshes exhibits less shape distortion and improved  ...  PIFu for the first time demonstrated high-quality single-view mesh reconstruction for clothed human with rich surface details, such as clothes wrinkles.  ... 
arXiv:2006.08072v2 fatcat:7qbnswptdfapzh2e5bxygvrsgy

Ray-ONet: Efficient 3D Reconstruction From A Single RGB Image [article]

Wenjing Bian and Zirui Wang and Kejie Li and Victor Adrian Prisacariu
2021 arXiv   pre-print
We propose Ray-ONet to reconstruct detailed 3D models from monocular images efficiently.  ...  Networks (ONet), while reducing the network inference complexity to O(N^2).  ...  Acknowledgements We thank Theo Costain for helpful discussions and comments. We thank Stefan Popov for providing the code for CoReNet and guidance on training.  ... 
arXiv:2107.01899v2 fatcat:yd6tjvaoe5aghmzqttagwzn7ke

ANISE: Assembly-based Neural Implicit Surface rEconstruction [article]

Dmitry Petrov, Matheus Gadelha, Radomir Mech, Evangelos Kalogerakis
2022 arXiv   pre-print
We present our results in well-known sparse point cloud reconstruction and single-view reconstruction benchmarks.  ...  We present ANISE, a method that reconstructs a 3D shape from partial observations (images or sparse point clouds) using a part-aware neural implicit shape representation.  ...  We introduce a shape representation for 3D reconstruction through a set assembly of neural implicits (left). Our approach results in state-of-the-art reconstruction quality.  ... 
arXiv:2205.13682v1 fatcat:73xrvfyeeffd3ijkpo3vckttii

DR-KFS: A Differentiable Visual Similarity Metric for 3D Shape Reconstruction [article]

Jiongchao Jin, Akshay Gadi Patil, Zhang Xiong, Hao Zhang
2020 arXiv   pre-print
We introduce a differential visual similarity metric to train deep neural networks for 3D reconstruction, aimed at improving reconstruction quality.  ...  optimize the network weights to produce reconstructions with better structural fidelity and visual quality.  ...  We introduce a differential visual similarity metric to train deep neural networks for 3D reconstruction, aimed at improving reconstruction quality.  ... 
arXiv:1911.09204v4 fatcat:swc5tzdtfzdsrjv6dkq6lgjqgm
« Previous Showing results 1 — 15 out of 41 results