A deep learning approach to the classification of 3D CAD models

Qin, Feiwei; Lu-ye, LI; Gao, Shiming; Yang, Xiaoling; Chen, Xiang

doi:10.1631/jzus.c1300185

Cited by 60 publications

(30 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In this domain, Qin et al (2014) IfcProductDefinitionShape, which contain detailed information about the beam. By tracing the references between entities in an IFC data, all relevant information to an object can be extracted (Won et al 2013).…”

Section: Object Classificationmentioning

confidence: 99%

New Automated BIM Object Classification Method to Support BIM Interoperability

Zhang

2019

J. Comput. Civ. Eng.

View full text Add to dashboard Cite

Industry foundation classes (IFC) is widely accepted as the future of building information modeling (BIM) to take on the challenge of BIM interoperability and enables its support of various automation tasks. However, it is not uncommon to see misuses of IFC entities during the creation of BIM. Such misuses prevent a successful automation of BIM-supported tasks because misclassification of objects in BIM can lead to significant negative consequences in downstream applications due to incorrect semantic information provided. To address this problem, the authors propose a new data-driven, iterative method that can be used to develop an algorithm to automatically classify each object in an IFC model into predefined categories. The algorithm consists of multiple sub-algorithms with each sub-algorithm depicting a pattern matching rule that uses inherent features of the geometric representation of an architecture, engineering, and construction (AEC) object. The method was tested in an experiment where IFC models from three different sources were collected and 1,891 AEC objects were extracted and divided into training and testing data for use. By comparing the classification results of the algorithm developed based on training data and applied to testing data with a manually developed gold standard, 84.45% recall and 85.20% precision were achieved in common building element categories, 100% recall and 2 precision were achieved in detailed beam categories. The sources of errors were found to be: (1) different objects sharing the same geometric shape; and (2) uncovered geometric shape representation in the training data. By adding locational information into consideration in addition to geometric information and making sure training data covers all geometric shape representations, 100% precision and recall can be achieved for all categories.

show abstract

Section: Object Classificationmentioning

confidence: 99%

New Automated BIM Object Classification Method to Support BIM Interoperability

Zhang

2019

J. Comput. Civ. Eng.

View full text Add to dashboard Cite

show abstract

“…Most existing shape classification methods usually use the labeled model set as the training data, and train a classifier based on the supervised learning methods, such as nearest neighbor classifier (Csakany and Wallace, 2003;Donamukkala et al, 2005;Biasotti et al, 2006), Bayesian classifier (Huber et al, 2004), SVM (Marini et al, 2011;Barra and Biasotti, 2014), belief function (Tabia et al, 2013) and deep neural network classifier (Qin et al, 2014). A recent work presented by Huang et al (2013b) uses a semi-supervised method for fine-grained 3D shape classification with several pre-labeled samples.…”

Section: D Shape Classificationmentioning

confidence: 99%

“…Most existing works (Csakany and Wallace, 2003;Biasotti et al, 2006;Huber et al, 2004;Tabia et al, 2011Tabia et al, , 2013Marini et al, 2011;Barra and Biasotti, 2014;Qin et al, 2014) have focused on learning a classifier by a large number of labeled samples to classify a given shape. As the training set utterly determines the scope of categories and the trained classifier is hardly updated, almost all of them cannot generalize well to unknown object category and multiple classification criterions to fit the needs of the diversity of taxonomies and the different application requirements.…”

Section: Introductionmentioning

confidence: 99%

Iterative 3D shape classification by online metric learning

Song

Sun

Liu

et al. 2015

Computer Aided Geometric Design

View full text Add to dashboard Cite

“…The high-level abstraction representation can be obtained from a DBM which is trained by depth images of 3D model and the feature is used in semisupervised classification method. Qin et al [29] proposed a deep learning approach to automatically classify 3D CAD models according to the mechanical part catalogue. The designed deep neural network classifier is based on the latest machine learning technique, deep learning.…”

Section: Introductionmentioning

confidence: 99%

A Novel 3D CAD Model Retrieval Method Based on Vertices Classification and Weights Combination Optimization

Zhuang

Zhang

Hou

et al. 2017

Mathematical Problems in Engineering

View full text Add to dashboard Cite

3D shape retrieval is a problem of current interest in different fields, especially in the mechanical engineering domain. According to our knowledge, multifeature based techniques achieve the best performance at present. However, the practicability of those methods is badly limited due to the high computational cost. To improve the retrieval efficiency of 3D CAD model, we propose a novel 3D CAD model retrieval algorithm called VSC_WCO which consists of a new 3D shape descriptor named VSC and Weights Combination Optimization scheme WCO. VSC represents a 3D model with three distance distribution histograms based on vertices classification. The weighted sum of L1 norm distances between corresponding distance histograms of two VSC descriptors is regarded as dissimilarity of two models. For higher retrieval accuracy on a classified 3D model database, WCO is proposed based on Particle Swarm Optimization and existing class information. Experimental results on ESB, PSB, and NTU databases show that the discriminative power of VSC is already comparable to or better than several typical shape descriptors. After WCO is employed, the performance of VSC_WCO is similar to the leading methods by all performance metrics and is much better by computational efficiency.

show abstract

A deep learning approach to the classification of 3D CAD models

Cited by 60 publications

References 20 publications

New Automated BIM Object Classification Method to Support BIM Interoperability

New Automated BIM Object Classification Method to Support BIM Interoperability

Iterative 3D shape classification by online metric learning

A Novel 3D CAD Model Retrieval Method Based on Vertices Classification and Weights Combination Optimization

Contact Info

Product

Resources

About