CAD Priors for Accurate and Flexible Instance Reconstruction

@article{Birdal2017CADPF,
  title={CAD Priors for Accurate and Flexible Instance Reconstruction},
  author={Tolga Birdal and Slobodan Ilic},
  journal={2017 IEEE International Conference on Computer Vision (ICCV)},
  year={2017},
  pages={133-142}
}
We present an efficient and automatic approach for accurate instance reconstruction of big 3D objects from multiple, unorganized and unstructured point clouds, in presence of dynamic clutter and occlusions. In contrast to conventional scanning, where the background is assumed to be rather static, we aim at handling dynamic clutter where the background drastically changes during object scanning. Currently, it is tedious to solve this problem with available methods unless the object of interest… 

Figures and Tables from this paper

A Minimalist Approach to Type-Agnostic Detection of Quadrics in Point Clouds
TLDR
This paper proposes a segmentation-free, automatic and efficient procedure to detect general geometric quadric forms in point clouds, where clutter and occlusions are inevitable and results on synthetic and real datasets support the validity of the method.
Learning Multiview 3D Point Cloud Registration
TLDR
Experimental evaluation on well accepted benchmark datasets shows that the proposed end-to-end algorithm for joint learning of both parts of this two-stage registration problem outperforms the state-of-the-art by a significant margin, while being end- to-end trainable and computationally less costly.
A point sampling algorithm for 3D matching of irregular geometries
  • Tolga Birdal, Slobodan Ilic
  • Computer Science
    2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
  • 2017
TLDR
This work develops a simple, yet very effective point sampling strategy for detection of any CAD model through geometric hashing, which relies on rendering the object coordinates from a set of views evenly distributed on a sphere.
3D Local Features for Direct Pairwise Registration
TLDR
RelativeNet is introduced, a relative pose estimation network to assign correspondence-specific orientations to the keypoints, eliminating any local reference frame computations and suggesting that augmenting the key points with local pose information leads to better generalization and a dramatic speed-up.
End-to-end globally consistent registration of multiple point clouds
TLDR
Experimental evaluation on benchmark datasets shows that the proposed end-to-end algorithm for joint learning of both parts of this two-stage problem outperforms state-of-the-art by a significant margin, while being end- to-end trainable and computationally less costly.
From Planes to Corners: Multi-Purpose Primitive Detection in Unorganized 3D Point Clouds
TLDR
This work proposes a new method for segmentation-free joint estimation of orthogonal planes, their intersection lines, relationship graph and corners lying at the intersection of three orthogonality planes, and forms a graph of these primitives, paving the way to the extraction of further reliable features: lines and corners.
MultiBodySync: Multi-Body Segmentation and Motion Estimation via 3D Scan Synchronization
TLDR
This work presents MultiBodySync, a novel, end-to-end trainable multi-body motion segmentation and rigid registration framework for multiple input 3D point clouds that incorporates spectral synchronization into an iterative deep declarative network so as to simultaneously recover consistent correspondences as well asmotion segmentation.
Deep Bingham Networks: Dealing with Uncertainty and Ambiguity in Pose Estimation
TLDR
This work introduces Deep Bingham Networks (DBN), a generic framework that can naturally handle pose-related uncertainties and ambiguities arising in almost all real life applications concerning 3D data and proposed new training strategies so as to avoid mode or posterior collapse during training and to improve numerical results.
...
...

References

SHOWING 1-10 OF 55 REFERENCES
Dense Object Reconstruction with Semantic Priors
TLDR
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 and can produce more accurate reconstructions than alternative state-of-the-art multiv view stereo systems.
SLAM++: Simultaneous Localisation and Mapping at the Level of Objects
TLDR
The object graph enables predictions for accurate ICP-based camera to model tracking at each live frame, and efficient active search for new objects in currently undescribed image regions, as well as the generation of an object level scene description with the potential to enable interaction.
Multi-view object class detection with a 3D geometric model
  • Joerg Liebelt, C. Schmid
  • Computer Science
    2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition
  • 2010
TLDR
This paper presents a new approach for multi-view object class detection which uses a part model which discriminatively learns the object appearance with spatial pyramids from a database of real images, and encodes the 3D geometry of the object class with a generative representation built from adatabase of synthetic models.
Detailed 3D Representations for Object Recognition and Modeling
TLDR
This paper revisit ideas from the early days of computer vision, namely, detailed, 3D geometric object class representations for recognition, and demonstrates novel applications enabled by such an object class representation, such as fine-grained categorization of cars and bicycles, according to their 3D geometry, and ultrawide baseline matching.
Rigid 3D geometry matching for grasping of known objects in cluttered scenes
TLDR
This application demonstrates how the integration of computer vision and soft-robotics leads to a robotic system capable of acting in unstructured and occluded environments.
Smooth Signed Distance Surface Reconstruction and Applications
TLDR
A new and simple variational formulation to reconstruct the surface geometry, topology, and color map of a 3D scene from a finite set of colored oriented points is described, which produces high quality polygon meshes with smooth color maps, which accurately approximate the sourcecolored oriented points.
SDF-2-SDF: Highly Accurate 3D Object Reconstruction
TLDR
This work proposes a novel implicit-to-implicit surface registration scheme between signed distance fields (SDFs) that circumvents expensive correspondence search and allows for incorporation of multiple geometric constraints without any dependence on texture, yielding highly accurate 3D models.
Point Pair Features Based Object Detection and Pose Estimation Revisited
TLDR
A revised pipe-line of the existing 3D object detection and pose estimation framework based on point pair feature matching is presented, and it is argued that such a combined pipeline simultaneously boosts the detection rate and reduces the complexity, while improving the accuracy of the resulting pose.
SDF Tracker: A parallel algorithm for on-line pose estimation and scene reconstruction from depth images
TLDR
This work proposes a simple on-line method for tracking the pose of a depth camera in six degrees of freedom and simultaneously maintaining an updated 3D map, represented as a truncated signed distance function.
Three-Dimensional Model-Based Object Recognition and Segmentation in Cluttered Scenes
TLDR
A novel 3D model-based algorithm is presented which performs viewpoint independent recognition of free-form objects and their segmentation in the presence of clutter and occlusions automatically and efficiently and is superior in terms of recognition rate and efficiency.
...
...