Abstract
Point clouds provide an efficient way for 3D geometric object representation. In order to deal with the classification and segmentation of point cloud, it is very important to design an efficient and intelligent model that can directly affect point cloud. Due to the irregularity of the data format, traditional convolutional neural networks cannot be applied to point clouds processing directly. Graph convolution network (GCN) has attracted more and more attention in recent years, especially in the field of non-Euclidean data processing. Point clouds processing with GCN models is an efficient and suitable method, a lot of GCN models have achieved state-of-the-art performance on irregular data processing challenges. In this paper, we propose a Multi-scale Dynamic GCN model for point clouds classification, a Farthest Point Sampling method is applied in our model firstly to efficiently cover the entire point set, it uses different scale k-NN group method to locate on k nearest neighborhood for each central node, Edge Convolution (EdgeConv) operation is used to extract and aggregate local features between neighbor connected nodes and central node. We use ModelNet40, ModelNet10 and ShapeNet part dataset to classify point clouds and segment them semantically. Experiments show that our model achieves a better performance on classification accuracy and model complexity than other state-of-the-art models.
Highlights
Point clouds are very important spatial geometric data, they provide a flexible and scalable geometry representation suitable for many applications in computer graphics
Some latest works [8]–[12] were employing the same data processing method as PointNet, directly took raw point clouds as input without any converting them to other formats, PointNet++ [13] added a local adjacent neighborhood concept to capture more local hierarchical features during the feature extraction process and applied a global max pooling layer to aggregate global semantic information as PointNet dose; DGCNN [14] could dynamically update the graph, it applied an edge convolution (EdgeConv) operation to capture the local relationship between adjacent nodes and central node, used the connected edge between adjacent node and central node to represent the synthesis features of this two nodes, utilized an Multi-Layer Perception (MLP)
DATASET We evaluated our 3D point clouds semantic segmentation model on ShapeNet part dataset, the dataset contains 16881 3D shapes from 16 object categories, annotated with 50 parts in total, 12137 models were used for training, 1870 models are used for validation and 2874 models are for testing. 2048 points were sampled from each training shape at the beginning, and most sampled point set are labeled with less than 6 parts
Summary
Point clouds are very important spatial geometric data, they provide a flexible and scalable geometry representation suitable for many applications in computer graphics. Since PointNet has proved that processing the original point clouds data directly would have a better performance than preprocessing with manual feature engineering methods, so more and more similar neural network models are proposed after PointNet. Some latest works [8]–[12] were employing the same data processing method as PointNet, directly took raw point clouds as input without any converting them to other formats, PointNet++ [13] added a local adjacent neighborhood concept to capture more local hierarchical features during the feature extraction process and applied a global max pooling layer to aggregate global semantic information as PointNet dose; DGCNN [14] could dynamically update the graph, it applied an edge convolution (EdgeConv) operation to capture the local relationship between adjacent nodes and central node, used the connected edge between adjacent node and central node to represent the synthesis features of this two nodes, utilized an MLP. The goal is to select L starting points as the center point of the step
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
