Fine-grained semi-supervised labeling of large shape collections

By Qi-Xing Huang, Hao Su, Leonidas J. Guibas. 2013

79 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Industrial and Manufacturing Engineering

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Industrial and Manufacturing Engineering

Citations in the corpus

SyncSpecCNN: Synchronized Spectral CNN for 3D Shape Segmentation (2016, methodology): “[10] first jointly aligns all the shapes in 3D space via free-form deformation and then propagates keypoint labels to test shapes from its K nearest training shapes.”
ShapeNet: An Information-Rich 3D Model Repository (2015, background): “• Rigid Alignments: Establishing a consistent canonical orientation (e.g., upright and front) for every model is important for various tasks such as visualizing shapes [13], shape classiﬁcation [8] and shape recognition [34].”
Multi-view 3D Models from Single Images with a Convolutional Network (2015, methodology): “The models have been semi-automatically aligned to a consistent orientation using a hierarchical approach based on [25].”
Joint embeddings of shapes and images via CNN image purification (2015, methodology): “The instances in the set are jointly aligned by a global rigid transformation [Huang et al. 2013], and then projected from k viewpoints to generate projection images: Ii = {Ii,v}kv=1 for each Si.”
A survey of deep neural network architectures and their applications (2017, background): “Previously, the mesh labeling approaches fo cused on the mesh triangle which was characterized by heuristically designed geometry features [72], [80].”
Learning class‐specific descriptors for deformable shapes using localized spectral convolutional networks (2015, unknown): “…processing trend of learning invariant structure rather than trying to hand-craft them, several learning frameworks have been proposed in the geometry processing community as well, for applications such as correspondence [RRBW ∗ 14], retrieval [LBBC14], labelling and segmentation [KHS10, HSG13].”
Deformation-driven shape correspondence via shape recognition (2017, background): “This would be possible if a ne-grained shape categorization [Huang et al. 2013] can be obtained and utilized.”
A scalable active framework for region annotation in 3D shape collections (2016, methodology): “In particular, we use the hierarchical joint alignment algorithm of Chang et al. [2015], and then refine the alignments with free-form deformations [Huang et al. 2013].”
Elements of style (2015, methodology): “A range of methods provide strategies for evaluating fine-grained similarity between shapes with similar structure [Xu et al. 2010; Kalogerakis et al. 2012; Kim et al. 2013; Huang et al. 2013; van Kaick et al. 2013; Yumer and Kara 2014].”
3D Object Classification via Spherical Projections (2017, background): “While early works focus on leveraging handcrafted features for classification [14, 1, 12, 8, 15, 9], recent methods seek to leverage the power of deep neural networks [27, 25].”
Estimating image depth using shape collections (2014, methodology): “Recently, we have witnessed the success of data-driven techniques in shape analysis [Huang et al. 2011; Kim et al. 2012a; Kim et al. 2013; Huang et al. 2013; Wang et al. 2013], shape model- ing [Chaudhuri et al. 2011; Kalogerakis et al. 2012; Averkiou et al. 2014] and shape reconstruction [Nan et…”
Shape2Pose (2014, methodology): “In particular, we use the method of [Huang et al. 2013] to co-align all shapes in the entire dataset.”