Single-view reconstruction via joint analysis of image and shape collections

By Qi-Xing Huang, Hai Wang, Vladlen Koltun. 2015

153 citations recorded by the source corpus.

Subfields: Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition, 3D Vision & Pose

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Graphics and Computer-Aided Design

In Computer Vision and Pattern Recognition

In 3D Vision & Pose

Citations in the corpus

A Point Set Generation Network for 3D Object Reconstruction from a Single Image (2016, background): “For example, [19, 13] proposed to deform and reassemble existing shapes into a new model to fit the observed image.”
RealPoint3D: Point Cloud Generation from a Single Image with Complex Background (2018, background): “In (Su et al. 2014) and (Huang, Wang, and Koltun 2015), the authors reconstructed the depth of objects from web-site collected images by exploiting a collection of aligned 3D models of related objects shape.”
OctNet: Learning Deep 3D Representations at High Resolutions (2016, background): “However, recent advances in 3D reconstruction [33] and graphics [21] allow capturing and modeling large amounts of 3D data.”
Learning Shape Priors for Single-View 3D Completion and Reconstruction (2018, background): “[21] propose to borrows shape parts from existing CAD models.”
Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images (2018, background): “Huang et al. [15] and Su et al. [29] retrieve shape components from a large dataset, assemble them and deform the assembled shape to ﬁt the observed image.”
Learning a Predictable and Generative Vector Representation for Objects (2016, methodology): “Many approaches achieve this via alignment [1, 14, 22]; the most similar work is Li et al.”
Single Image 3D Interpreter Network (2016, methodology): “Depth-/mesh-based models can recover detailed 3D object structure from a single image, either by adapting existing 3D models from a database [3,19,6,8,20,21,22,23,24], or by inferring from its detected 2D silhouette [5,7,4].”
Multi-view 3D Models from Single Images with a Convolutional Network (2015, background): “Huang et al. [21] reconstruct 3D models from single images of objects by jointly analyzing large collections of images and 3D models of objects of the same kind.”
Marr Revisited: 2D-3D Alignment via Surface Normal Prediction (2016, background): “Recent work include instance detection of a small set of IKEA objects via contour-based alignment [26], depth prediction by aligning to renders of 3D shapes via hand-crafted features [44], object class detection via exemplar matching with mid-level elements [1, 7], and alignment via composition from”
Learning 3D Object Categories by Looking Around Them (2017, methodology): “Starting from datasets of unordered images, methods such as [44] and [27] use SFM and manual annotations, such as keypoints in [4, 16], to estimate a rough 3D geometry of objects.”
Joint embeddings of shapes and images via CNN image purification (2015, methodology): “The unified space also supports various other applications that require 3D representations of objects observed in images [Xu et al. 2011; Zheng et al. 2012; Su et al. 2014; Kholgade et al. 2014; Lee et al. 2015; Huang et al. 2015].”
Sparse Representation for 3D Shape Estimation: A Convex Relaxation Approach (2015, background): “Another line of work tried to solve single-image reconstruction by finding the nearest neighbor in the shape collection followed by a refinement step [23], [24].”
IM2CAD (2016, methodology): “Third, we leverage deep features trained by convolutional neural nets (CNNs) [21] to reliably match between photographs and CAD renderings [3, 20, 36, 18].”
Learning Dense Correspondence via 3D-Guided Cycle Consistency (2016, methodology): “obtaining direct supervision for Fr1,r2 through Fr1,s1◦F̃s1,s2◦Fs2,r2 ,where Fr1,s1 and Fs2,r2 are inferred from [17]).”
Cross-Domain Self-Supervised Multi-task Feature Learning Using Synthetic Imagery (2017, methodology): “Synthetic data for vision CAD models have been used for various vision tasks such as 2D-3D alignment [6, 4], object detection [53], joint pose estimation and image-shape alignment [64, 27].”
Learning Category-Specific Deformable 3D Models for Object Reconstruction (2017, background): “Index Terms—Object reconstruction, 3D shape modeling, viewpoint estimation, scene understanding Ç”
Configurable 3D Scene Synthesis and 2D Image Rendering with Per-pixel Ground Truth Using Stochastic Grammars (2017, unknown): “2016), single-view reconstruction (Huang et al. 2015), view-point estimation (Movshovitz-Attias et al.”
Pixel2Mesh++: Multi-View 3D Mesh Generation via Deformation (2019, unknown): “[16] and Su et al .”
Learning Shape Abstractions by Assembling Volumetric Primitives (2016, background): “In the domain of 3D shapes, the idea of exploiting compositionality has played a similarly important role for object representation, parsing, and manipulation.”