A scalable active framework for region annotation in 3D shape collections

By Li Yi, Vladimir G. Kim, Duygu Ceylan, I-Chao Shen, Mengyan Yan, Hao Su, Cewu Lu, Qi-Xing Huang, Alla Sheffer, Leonidas J. Guibas. 2016

1,414 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Geology

How researchers see this paper

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

In Computational Mechanics

In Computer Vision and Pattern Recognition

In Geology

Citations in the corpus

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation (2016, methodology): “We evaluate on ShapeNet part data set from [26], which contains 16,881 shapes from 16 categories, annotated with 50 parts in total.”
PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space (2017, methodology): “Following the setting in [32], we evaluate our approach on part segmentation task assuming category label for each shape is already known.”
SyncSpecCNN: Synchronized Spectral CNN for 3D Shape Segmentation (2016, methodology): “We compare our framework with traditional learning-based techniques [27, 29] leveraging on local geometric features and shape alignment cues, as well as recent deep learning based approaches [2] which also fall into the family of graph CNNs.”
PointCNN: Convolution On X-Transformed Points (2018, methodology): “We show detailed segmentation result comparisons on ShapeNet Parts in Table 1, we can see our approach achieves the best overall performance and are best on 7 of the 16 categories.”
Local Spectral Graph Convolution for Point Set Feature Learning (2018, methodology): “We use the ofﬁcial training/testing split, following [11,20,21], where the challenge is to assign a part label to each point in the test set.”
Vector Neurons: A General Framework for SO(3)-Equivariant Networks (2021, methodology): “For the ShapeNet dataset, we followed [38] by using ShapeNet-part for part segmentation, which has 16 shape categories with more than 30,000 models.”
Neighbors Do Help: Deeply Exploiting Local Structures of Point Clouds (2017, methodology): “We evaluate our work for part segmentation on ShapeNet part dataset [33].”
PAConv: Position Adaptive Convolution with Dynamic Kernel Assembling on Point Clouds (2021, methodology): “PAConv is also evaluated on ShapeNet Parts [61] for shape part segmentation.”
SO-Net: Self-Organizing Network for Point Cloud Analysis (2018, methodology): “Object part segmentation is demonstrated with the ShapeNetPart dataset [38].”
3D Shape Segmentation with Projective Convolutional Networks (2016, methodology): “The dataset from ShapeNetCore currently contains 17,773 expert-verified segmentations of 3D models across 16 categories.”
Multiresolution Tree Networks for 3D Point Cloud Processing (2018, methodology): “Previous work has also studied extensions of ConvNets to mesh surfaces such as spectral CNNs [5,45], geodesic CNNs [29], or anisotropic CNNs [2].”
GSPN: Generative Shape Proposal Network for 3D Instance Segmentation in Point Cloud (2018, methodology): “Different from previous large-scale part annotation dataset [46], PartNet provides ground truth part instance annotation in addition to semantic labeling and the segmentation granularity is more detailed, making it more suitable for testing instance segmentation approaches.”
Recurrent Slice Networks for 3D Segmentation of Point Clouds (2018, methodology): “Another one is the ShapeNet dataset [35], a synthetic dataset.”
Escape from Cells: Deep Kd-Networks for the Recognition of 3D Point Cloud Models (2017, methodology): “Our results are compared to 3D-CNN (reproduced from [20]), PointNet architecture [20], and the architecture of [37].”
LPD-Net: 3D Point Cloud Learning for Large-Scale Place Recognition and Environment Analysis (2018, background): “ModelNet [26] and ShapeNet [27]), but this is not the case for large-scale environments, since such point clouds are mainly composed of different objects in the scene and with unknown relationships between the objects.”
Point-M2AE: Multi-scale Masked Autoencoders for Hierarchical Point Cloud Pre-training (2022, methodology): “We evaluate Point-M2AE for part segmentation on ShapeNetPart [48], which predicts per-point part labels and requires detailed understanding for local patterns.”
StarMap for Category-Agnostic Keypoint and Viewpoint Estimation (2018, methodology): “Also, annotating 3D keypoints of a CAD model [47] is not a hard task, given an interactive 3D UI such as MeshLab [5].”