Full Text:  <3284>

Summary:  <580>

CLC number: TP391.72

On-line Access: 2014-01-29

Received: 2013-07-09

Revision Accepted: 2013-11-21

Crosschecked: 2014-01-15

Cited: 5

Clicked: 2247

Citations:  Bibtex RefMan EndNote GB/T7714

-   Go to

Article info.
1. Reference List
Open peer comments

Journal of Zhejiang University SCIENCE C 2014 Vol.15 No.2 P.91-106

10.1631/jzus.C1300185


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


Author(s):  Fei-wei Qin, Lu-ye Li, Shu-ming Gao, Xiao-ling Yang, Xiang Chen

Affiliation(s):  State Key Lab of CAD & CG, Zhejiang University, Hangzhou 310058, China

Corresponding email(s):   qinfeiwei@zjucadcg.cn, liluye@cad.zju.edu.cn, smgao@cad.zju.edu.cn, xchen@cad.zju.edu.cn, sunny_aday@163.com

Key Words:  CAD model classification, Design reuse, Machine learning, Neural network


Fei-wei Qin, Lu-ye Li, Shu-ming Gao, Xiao-ling Yang, Xiang Chen. A deep learning approach to the classification of 3D CAD models[J]. Journal of Zhejiang University Science C, 2014, 15(2): 91-106.

@article{title="A deep learning approach to the classification of 3D CAD models",
author="Fei-wei Qin, Lu-ye Li, Shu-ming Gao, Xiao-ling Yang, Xiang Chen",
journal="Journal of Zhejiang University Science C",
volume="15",
number="2",
pages="91-106",
year="2014",
publisher="Zhejiang University Press & Springer",
doi="10.1631/jzus.C1300185"
}

%0 Journal Article
%T A deep learning approach to the classification of 3D CAD models
%A Fei-wei Qin
%A Lu-ye Li
%A Shu-ming Gao
%A Xiao-ling Yang
%A Xiang Chen
%J Journal of Zhejiang University SCIENCE C
%V 15
%N 2
%P 91-106
%@ 1869-1951
%D 2014
%I Zhejiang University Press & Springer

TY - JOUR
T1 - A deep learning approach to the classification of 3D CAD models
A1 - Fei-wei Qin
A1 - Lu-ye Li
A1 - Shu-ming Gao
A1 - Xiao-ling Yang
A1 - Xiang Chen
J0 - Journal of Zhejiang University Science C
VL - 15
IS - 2
SP - 91
EP - 106
%@ 1869-1951
Y1 - 2014
PB - Zhejiang University Press & Springer
ER -


Abstract: 
Model classification is essential to the management and reuse of 3D CAD models. Manual model classification is laborious and error prone. At the same time, the automatic classification methods are scarce due to the intrinsic complexity of 3D CAD models. In this paper, we propose an automatic 3D CAD model classification approach based on deep neural networks. According to prior knowledge of the CAD domain, features are selected and extracted from 3D CAD models first, and then preprocessed as high dimensional input vectors for category recognition. By analogy with the thinking process of engineers, a deep neural network classifier for 3D CAD models is constructed with the aid of deep learning techniques. To obtain an optimal solution, multiple strategies are appropriately chosen and applied in the training phase, which makes our classifier achieve better performance. We demonstrate the efficiency and effectiveness of our approach through experiments on 3D CAD model datasets.

用于三维CAD模型分类的深度学习方法

研究目的:模型分类对于工程领域中三维CAD模型的有效管理和重用非常关键。由于三维CAD模型分类固有的复杂性,难以通过设计一定的规则实现模型自动分类。因此,一些学者采用机器学习技术解决三维CAD模型的自动分类问题。受传统机器学习技术的制约,已有方法效果并不理想,无法应用于实际工业生产。近年来,深度学习取得了突破性进展,在图像分类、自然语言处理、语音识别等领域获得了不错效果。本文探索利用机器学习领域的最新研究成果——深度学习技术,对工程领域中常见的三维CAD模型进行自动分类。
创新要点:利用深度学习技术,模拟工程师人工识别三维零件的三个主要环节,构建了CAD模型的自动分类器;采用多种行之有效的训练策略,得到具有较好泛化性能的分类器。
研究方法:基于深度学习的三维CAD模型分类方法的主要步骤为:从制造企业获取足够的样本数据,构建三维CAD模型数据集;分析CAD领域的特点,选取合适的三维模型描述符进行特征提取,并处理为输入向量;利用深度学习技术模拟工程师人工识别工程零件的过程,设计分类器的拓扑结构并选择相应的组成元素,构建深层神经网络分类器;采用多种实用的训练技巧,使训练得到的分类器的自由参数取值尽可能逼近全局最优解,避免过拟合。
重要结论:实验数据表明,将深度学习技术与CAD领域知识结合,能够取得满意效果。深层神经网络分类器在测试集上的分类误差率低,足以满足工业生产的需求。

关键词:CAD模型分类;设计重用;机器学习;神经网络

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

[1]Bai, J., Gao, S., Tang, W., et al., 2010. Design reuse oriented partial retrieval of CAD models. Comput.-Aided Des., 42(12):1069-1084.

[2]Barutcuoglu, Z., DeCoro, C., 2006. Hierarchical shape classification using Bayesian aggregation. IEEE Int. Conf. on Shape Modeling and Applications, p.44-48.

[3]Bengio, Y., 2009. Learning deep architectures for AI. Found. Trends Mach. Learn., 2(1):1-127.

[4]Bengio, Y., Courville, A., Vincent, P., 2013. Representation learning: a review and new perspectives. IEEE Trans. Pattern Anal. Mach. Intell., 35(8):1798-1828.

[5]Bergstra, J., Breuleux, O., Bastien, F., et al., 2010. Theano: a CPU and GPU math compiler in Python. Proc. 9th Python in Science Conf., p.1-7.

[6]Bimbo, A.D., Pala, P., 2006. Content-based retrieval of 3D models. ACM Trans. Multim. Comput. Commun. Appl., 2(1):20-43.

[7]Bishop, C.M., 1995. Neural Networks for Pattern Recognition. Oxford University Press, Oxford.

[8]Bordes, A., Glorot, X., Weston, J., et al., 2014. A semantic matching energy function for learning with multi-relational data. Mach. Learn., 94(2):233-259.

[9]Chen, D., 2003. Three-Dimensional Model Shape Description and Retrieval Based on Light Field Descriptors. PhD Thesis, National Taiwan University, Taiwan.

[10]Chen, D., Tian, X., Shen, Y., et al., 2003. On visual similarity based 3D model retrieval. Comput. Graph. Forum, 22(3):223-232.

[11]Dreiseitl, S., Ohno-Machado, L., 2002. Logistic regression and artificial neural network classification models: a methodology review. J. Biomed. Inform., 35(5):352-359.

[12]Duda, R.O., Hart, P.E., Stork, D.G., 2001. Pattern Classification (2nd Ed.). John Wiley & Sons, New York.

[13]Fang, X., Luo, H., Tang, J., 2005. Structural damage detection using neural network with learning rate improvement. Comput. & Struct., 83(25-26):2150-2161.

[14]Glorot, X., Bengio, Y., 2010. Understanding the difficulty of training deep feedforward neural networks. Proc. Int. Conf. on Artificial Intelligence and Statistics, p.249-256.

[15]Gunn, T.G., 1982. The mechanization of design and manufacturing. Sci. Am., 247:114-130.

[16]Haykin, S., 2008. Neural Networks and Learning Machines (3rd Ed.). Prentice Hall, New York.

[17]Hinton, G.E., Salakhutdinov, R.R., 2006. Reducing the dimensionality of data with neural networks. Science, 313(5786):504-507.

[18]Hou, S., Lou, K., Ramani, K., 2005. SVM-based semantic clustering and retrieval of a 3D model database. Comput. Aided Des. Appl., 2(1-4):155-164.

[19]Huang, F.J., LeCun, Y., 2006. Large-scale learning with SVM and convolutional nets for generic object categorization. IEEE Computer Society Conf. on Computer Vision and Pattern Recognition, p.284-291.

[20]Ip, C.Y., Regli, W.C., 2005a. Content-based classification of CAD models with supervised learning. Comput. Aided Des. Appl., 2(5):609-617.

[21]Ip, C.Y., Regli, W.C., 2005b. Manufacturing classification of CAD models using curvature and SVMs. Int. Conf. on Shape Modeling and Applications, p.361-365.

[22]Ip, C.Y., Regli, W.C., Sieger, L., et al., 2003. Automated learning of model classifications. Proc. 8th ACM Symp. on Solid Modeling and Applications, p.322-327.

[23]Iyer, N., Jayanti, S., Lou, K., et al., 2005. Three-dimensional shape searching: state-of-the-art review and future trends. Comput.-Aided Des., 37(5):509-530.

[24]Kavukcuoglu, K., Sermanet, P., Boureau, Y., et al., 2010. Learning convolutional feature hierarchies for visual recognition. Proc. 24th Annual Conf. on Neural Information Processing Systems, p.1090-1098.

[25]Krizhevsky, A., Sutskever, I., Hinton, G.E., 2012. ImageNet classification with deep convolutional neural networks. Proc. 26th Annual Conf. on Neural Information Processing Systems, p.1106-1114.

[26]Larochelle, H., Bengio, Y., Louradour, J., et al., 2009. Exploring strategies for training deep neural networks. J. Mach. Learn. Res., 10(1):1-40.

[27]Ngiam, J., Chen, Z., Koh, P.W., et al., 2011. Learning deep energy models. Proc. 28th Int. Conf. on Machine Learning, p.1105-1112.

[28]Prechelt, L., 1998. Automatic early stopping using cross validation: quantifying the criteria. Neur. Networks, 11(4):761-767.

[29]Ranzato, M.A., Mnih, V., Hinton, G.E., 2010. Generating more realistic images using gated MRF’s. Proc. 24th Annual Conference on Neural Information Processing Systems, p.2002-2010.

[30]Reed, R., 1993. Pruning algorithms—a survey. IEEE Trans. Neur. Networks, 4(5):740-747.

[31]Shilane, P., Min, P., Kazhdan, M., et al., 2004. The Princeton Shape Benchmark. Proc. Conf. on Shape Modeling Applications, p.167-178.

[32]Wang, W., Liu, X., Liu, L., 2013. Shape matching and retrieval based on multiple feature descriptors. Comput. Aided Draft. Des. Manuf., 23(1):60-67.

[33]Wei, W., Yang, Y., Lin, J., et al., 2008. Color-based 3D model classification using Hopfield neural network. Proc. Int. Conf. on Computer Science and Software Engineering, p.883-886.

[34]Wu, M.C., Jen, S.R., 1996. A neural network approach to the classification of 3D prismatic parts. Int. J. Adv. Manuf. Technol., 11(5):325-335.

[35]Yao, Y., Rosasco, L., Caponnetto, A., 2007. On early stopping in gradient descent learning. Constr. Approx., 26(2):289-315.

[36]Zhang, D., Lu, G., 2002. An integrated approach to shape based image retrieval. Proc. 5th Asian Conf. on Computer Vision, p.652-657.

Open peer comments: Debate/Discuss/Question/Opinion

<1>

editor@No address<No mail>

2014-01-29 19:26:10

Reviewer: This paper proposed a method for automatically classifying CAD models based on deep neural networks. It combines light field descriptor (LFD) and Zernike moments descriptor to construct a deep network classifier then all trainable parameters for the 3D CAD model classifier are trained. Experimental results have shown the applicability of the proposed method.



This work is the first one to my knowledge to apply the technique of deep learning in the classification of 3D models. The algorithm is technically sound and the results are good.



Deep learning has been widely studied and used in image community. I am curious why this technique is used in 3D content until now. And it is nice to see that this technique does work well for 3D shape classification.

Please provide your name, email address and a comment





Journal of Zhejiang University-SCIENCE, 38 Zheda Road, Hangzhou 310027, China
Tel: +86-571-87952276; Fax: +86-571-87952331; E-mail: jzus@zju.edu.cn
Copyright © 2000 - Journal of Zhejiang University-SCIENCE