Multi-resolution 3D CNN for learning multi-scale spatial features in CAD models

2021 ◽  
Vol 91 ◽  
pp. 102038
Author(s):  
Sambit Ghadai ◽  
Xian Yeow Lee ◽  
Aditya Balu ◽  
Soumik Sarkar ◽  
Adarsh Krishnamurthy
Keyword(s):  
Multi Scale ◽  
Spatial Features ◽  
Cad Models ◽  
3D Cnn

scholarly journals Multi-Level 3D CNN for Learning Multi-Scale Spatial Features

2019 ◽  
Author(s):  
Sambit Ghadai ◽  
Xian Yeow Lee ◽  
Aditya Balu ◽  
Soumik Sarkar ◽  
Adarsh Krishnamurthy
Keyword(s):  
Multi Scale ◽  
Spatial Features ◽  
Multi Level ◽  
3D Cnn

scholarly journals Multi-scale and multi-representation CAD models reconciliation for knowledge synthesis

CIRP Annals ◽  
2020 ◽  
Vol 69 (1) ◽  
pp. 137-140
Author(s):  
Lionel Roucoules ◽  
Frédéric Demoly
Keyword(s):  
Knowledge Synthesis ◽  
Multi Scale ◽  
Cad Models

Orthogonal Distance Fields Representation for Machine-Learning Based Manufacturability Analysis

2020 ◽  
Author(s):  
Aditya Balu ◽  
Sambit Ghadai ◽  
Soumik Sarkar ◽  
Adarsh Krishnamurthy
Keyword(s):  
Machine Learning ◽  
Computer Aided Design ◽  
Boundary Representation ◽  
Engineering Product ◽  
Distance Fields ◽  
Computer Aided ◽  
Cad Models ◽  
Product Design Process ◽  
3D Cnn ◽  
Aided Design

Abstract Computer-aided Design for Manufacturing (DFM) systems play an essential role in reducing the time taken for product development by providing manufacturability feedback to the designer before the manufacturing phase. Traditionally, DFM rules are hand-crafted and used to accelerate the engineering product design process by integrating manufacturability analysis during design. Recently, the feasibility of using a machine learning-based DFM tool in intelligently applying the DFM rules have been studied. These tools use a voxelized representation of the design and then use a 3D-Convolutional Neural Network (3D-CNN), to provide manufacturability feedback. Although these frameworks work effectively, there are some limitations to the voxelized representation of the design. In this paper, we introduce a new representation of the computer-aided design (CAD) model using orthogonal distance fields (ODF). We provide a GPU-accelerated algorithm to convert standard boundary representation (B-rep) CAD models into ODF representation. Using the ODF representation, we build a machine learning framework, similar to earlier approaches, to create a machine learning-based DFM system to provide manufacturability feedback. As proof of concept, we apply this framework to assess the manufacturability of drilled holes. The framework has an accuracy of more than 84% correctly classifying the manufacturable and non-manufacturable models using the new representation.

Image segmentation of nasopharyngeal carcinoma using 3D CNN with long-range skip connection and multi-scale feature pyramid

2020 ◽  
Vol 24 (16) ◽  
pp. 12671-12680
Author(s):  
Feng Guo ◽  
Canghong Shi ◽  
Xiaojie Li ◽  
Xi Wu ◽  
Jiliu Zhou ◽  
...  
Keyword(s):  
Image Segmentation ◽  
Nasopharyngeal Carcinoma ◽  
Long Range ◽  
Scale Feature ◽  
Multi Scale ◽  
Feature Pyramid ◽  
3D Cnn

scholarly journals LARGE SCALE 3D POINT CLOUD MODELING FROM CAD DATABASE IN COMPLEX INDUSTRIAL ENVIRONMENTS

2020 ◽  
Vol XLIII-B2-2020 ◽  
pp. 391-398
Author(s):  
M. Corsia ◽  
T. Chabardès ◽  
H. Bouchiba ◽  
A. Serna
Keyword(s):  
Real World ◽  
Computer Aided Design ◽  
Point Cloud ◽  
Large Scale ◽  
Region Growing ◽  
Cad Model ◽  
3D Point Cloud ◽  
Multi Scale ◽  
Cad Models ◽  
Cloud Modeling

Abstract. In this paper, we present a method to build Computer Aided Design (CAD) representations of dense 3D point cloud scenes by queries in a large CAD model database. This method is applied to real world industrial scenes for infrastructure modeling. The proposed method firstly relies on a region growing algorithm based on novel edge detection method. This algorithm is able to produce geometrically coherent regions which can be agglomerated in order to extract the objects of interest of an industrial environment. Each segment is then processed to compute relevant keypoints and multi-scale features in order to be compared to all CAD models from the database. The best fitting model is estimated together with the rigid six degree of freedom (6 DOF) transformation for positioning the CAD model on the 3D scene. The proposed novel keypoints extractor achieves robust and repeatable results that captures both thin geometrical details and global shape of objects. Our new multi-scale descriptor stacks geometrical information around each keypoint at short and long range, allowing non-ambiguous matching for object recognition and positioning. We illustrate the efficiency of our method in a real-world application on 3D segmentation and modeling of electrical substations.

scholarly journals A Novel 2D-3D CNN with Spectral-Spatial Multi-Scale Feature Fusion for Hyperspectral Image Classification

2021 ◽  
Vol 13 (22) ◽  
pp. 4621
Author(s):  
Dongxu Liu ◽  
Guangliang Han ◽  
Peixun Liu ◽  
Hang Yang ◽  
Xinglong Sun ◽  
...  
Keyword(s):  
Feature Extraction ◽  
Image Classification ◽  
Classification Scheme ◽  
Hyperspectral Image ◽  
Feature Fusion ◽  
Spectral Feature ◽  
Hyperspectral Image Classification ◽  
Scale Feature ◽  
Multi Scale ◽  
3D Cnn

Multifarious hyperspectral image (HSI) classification methods based on convolutional neural networks (CNN) have been gradually proposed and achieve a promising classification performance. However, hyperspectral image classification still suffers from various challenges, including abundant redundant information, insufficient spectral-spatial representation, irregular class distribution, and so forth. To address these issues, we propose a novel 2D-3D CNN with spectral-spatial multi-scale feature fusion for hyperspectral image classification, which consists of two feature extraction streams, a feature fusion module as well as a classification scheme. First, we employ two diverse backbone modules for feature representation, that is, the spectral feature and the spatial feature extraction streams. The former utilizes a hierarchical feature extraction module to capture multi-scale spectral features, while the latter extracts multi-stage spatial features by introducing a multi-level fusion structure. With these network units, the category attribute information of HSI can be fully excavated. Then, to output more complete and robust information for classification, a multi-scale spectral-spatial-semantic feature fusion module is presented based on a Decomposition-Reconstruction structure. Last of all, we innovate a classification scheme to lift the classification accuracy. Experimental results on three public datasets demonstrate that the proposed method outperforms the state-of-the-art methods.

scholarly journals Car style-holon recognition in computer-aided design

2018 ◽  
Vol 6 (4) ◽  
pp. 719-738 ◽  
Author(s):  
Egon Ostrosi ◽  
Jean-Bernard Bluntzer ◽  
Zaifang Zhang ◽  
Josip Stjepandić
Keyword(s):  
Computer Aided Design ◽  
Characteristic Model ◽  
Multi Scale ◽  
Characteristic Lines ◽  
Computer Aided ◽  
Cad Models ◽  
Style Shift ◽  
Aided Design ◽  
Scale Design ◽  
Progressive Accumulation

Abstract Multi-scale design can presumably stimulate greater intelligence in computer-aided design (CAD). Using the style-holon concept, this paper proposes a computational approach to address multi-scale style recognition for automobiles. A style-holon is both a whole—it contains sub-styles of which it is composed—as well as a part of a broader style. In this paper, we first apply a variable precision rough set-based approach to car evaluation and ranking. Secondly, we extracted and subsequently computed the each car's characteristic lines from the CAD models. Finally, we identified style-holons using the property of a double-headed style-holon. A style-holon is necessarily included in a typical vertical arrangement with progressive accumulation and forms a nested hierarchical order called a holarchy of styles. We adopted an interactive cluster analysis to recognize style-holons. Our results demonstrate that car style depended on each brand's individual strategy: a car is a form endowed with some structural stability. The style-holon also demonstrated that the evolution of characteristic lines should preserve the property of functional homeostasis (the same functional states) as well as the property of homeorhesis (the same stable course of change). For many car companies, stable brand recognition is an important design specification. The proposed approach was used to analyse a set of car styles as well as to assist in the design of characteristic model lines. A designer can also use this approach to evaluate whether or not the strategic requirement—style alignment with the style-holon of benchmarked cars--is satisfied. Highlights A style-holon is double-headed: a part of a greater style that contains sub-styles. A car's characteristic lines preserve the properties of homeostasis and homeorhesis. The Chinese style offers a unique context to consider functionality of a whole style. Shift from functional to emotional performance demonstrated in Chinese car brands. Evaluates the strategic requirement of style alignment with the selected style-holon.

Sign in / Sign up

Export Citation Format

Share Document