Shape from Photographs: A Multi-view Stereo Pipeline

Hernández, C.; Vogiatzis, George

doi:10.1007/978-3-642-12848-6_11

Cited by 16 publications

(10 citation statements)

References 53 publications

(95 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, ongoing research efforts in fine-grained recognition and detection of object parts may also benefit our semantic reconstruction framework. In our future work, we seek to demonstrate our system in an MRF-based MVS framework like [18], since it provides the flexibility to combine our shape prior with silhouette information from object detectors like [12].…”

Section: Discussion and Future Workmentioning

confidence: 99%

“…Depth map-based methods seek a labeling from the space of pixels to a set of discrete depth labels [21]. Volumetric methods, on the other hand, seek a binary partitioning of 3D space into object and non-object [18,30]. We choose the patch-based system [15] for demonstration, but our framework can be generalized to other approaches too.…”

Section: Relation To Prior Workmentioning

confidence: 99%

See 1 more Smart Citation

Dense Object Reconstruction with Semantic Priors

Bao

Chandraker²,

Lin³

et al. 2013

2013 IEEE Conference on Computer Vision and Pattern Recognition

View full text Add to dashboard Cite

We present a dense reconstruction approach that overcomes the drawbacks of traditional multiview stereo by incorporating semantic information in the form of learned category-level shape priors and object detection. Given training data comprised of 3D scans and images of objects from various viewpoints, we learn a prior comprised of a mean shape and a set of weighted anchor points. The former captures the commonality of shapes across the category, while the latter encodes similarities between instances in the form of appearance and spatial consistency. We propose robust algorithms to match anchor points across instances that enable learning a mean shape for the category, even with large shape variations across instances. We model the shape of an object instance as a warped version of the category mean, along with instance-specific details. Given multiple images of an unseen instance, we collate information from 2D object detectors to align the structure from motion point cloud with the mean shape, which is subsequently warped and refined to approach the actual shape. Extensive experiments demonstrate that our model is general enough to learn semantic priors for different object categories, yet powerful enough to reconstruct individual shapes with large variations. Qualitative and quantitative evaluations show that our framework can produce more accurate reconstructions than alternative state-of-the-art multiview stereo systems.

show abstract

Section: Discussion and Future Workmentioning

confidence: 99%

Section: Relation To Prior Workmentioning

confidence: 99%

Dense Object Reconstruction with Semantic Priors

Bao

Chandraker²,

Lin³

et al. 2013

2013 IEEE Conference on Computer Vision and Pattern Recognition

View full text Add to dashboard Cite

show abstract

“…3DR has been explored in a large body of extant work in computer vision, for problems such as structure from motion [10,16] or multiview stereo [1,7,11,12,14,18,22] and at times even with single view images [6]. Ingenious work on "Shape-from-X" has utilized priors on natural images to infer geometric features, with "X" being shading, texture, specularity, shadow and so on [2,17,23,28,39].…”

Section: Introductionmentioning

confidence: 99%

DeformNet: Free-Form Deformation Network for 3D Shape Reconstruction from a Single Image

Kurenkov

Garg

et al. 2018

2018 IEEE Winter Conference on Applications of Computer Vision (WACV)

126

View full text Add to dashboard Cite

3D reconstruction from a single image is a key problem in multiple applications ranging from robotic manipulation to augmented reality. Prior methods have tackled this problem through generative models which predict 3D reconstructions as voxels or point clouds. However, these methods can be computationally expensive and miss fine details. We introduce a new differentiable layer for 3D data deformation and use it in DEFORMNET to learn a model for 3D reconstructionthrough-deformation. DEFORMNET takes an image input, searches the nearest shape template from a database, and deforms the template to match the query image. We evaluate our approach on the ShapeNet dataset and show that -(a) the Free-Form Deformation layer is a powerful new building block for Deep Learning models that manipulate 3D data (b) DEFORMNET uses this FFD layer combined with shape retrieval for smooth and detail-preserving 3D reconstruction of qualitatively plausible point clouds with respect to a single query image (c) compared to other state-of-the-art 3D reconstruction methods, DEFORMNET quantitatively matches or outperforms their benchmarks by significant margins. For more information, visit: https://deformnet-site.github.io/DeformNet-website/.

show abstract

“…These methods require camera centers to be collinear (equivalent to lateral motion). Generalizations of these approaches can be found in the multi-view stereo literature, where the aggregated cost is computed along the optical ray in a discretized volume [7,6].…”

Section: Introductionmentioning

confidence: 99%

A data-fusion approach to motion-stereo

Malapelle

Fusiello

Rossi

et al. 2016

Signal Processing: Image Communication

View full text Add to dashboard Cite

This paper introduces a novel method for performing motion-stereo, based on dynamic integration of depth (or its proxy) measures obtained by pairwise stereo matching of video frames. The focus is on the data fusion issue raised by the motion-stereo approach, which is solved within a Kalman filtering framework. Integration occurs along the temporal and spatial dimension, so that the final measure for a pixel results from the combination of measures of the same pixel in time and whose of its neighbors. The method has been validated on both synthetic and natural images, using the simplest stereo matching strategy and a range of di↵erent confidence measures, and has been compared to baseline and optimal strategies.

show abstract

Shape from Photographs: A Multi-view Stereo Pipeline

Cited by 16 publications

References 53 publications

Dense Object Reconstruction with Semantic Priors

Dense Object Reconstruction with Semantic Priors

DeformNet: Free-Form Deformation Network for 3D Shape Reconstruction from a Single Image

A data-fusion approach to motion-stereo

Contact Info

Product

Resources

About