3D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning Approach
We consider the recent challenges of 3D shape analysis based on a volumetric CNN that requires a huge computational power. This high-cost approach forces to reduce the volume resolutions when applying 3D CNN on volumetric data. In this context, we propose a multiorientation volumetric deep neural ne...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8964382/ |
| _version_ | 1818646447873064960 |
|---|---|
| author | A. A. M. Muzahid Wanggen Wan Ferdous Sohel Naimat Ullah Khan Ofelia Delfina Cervantes Villagomez Hidayat Ullah |
| author_facet | A. A. M. Muzahid Wanggen Wan Ferdous Sohel Naimat Ullah Khan Ofelia Delfina Cervantes Villagomez Hidayat Ullah |
| author_sort | A. A. M. Muzahid |
| collection | DOAJ |
| description | We consider the recent challenges of 3D shape analysis based on a volumetric CNN that requires a huge computational power. This high-cost approach forces to reduce the volume resolutions when applying 3D CNN on volumetric data. In this context, we propose a multiorientation volumetric deep neural network (MV-DNN) for 3D object classification with octree generating low-cost volumetric features. In comparison to conventional octree representations, we propose to limit the octree partition to a certain depth to reserve all leaf octants with sparsity features. This allows for improved learning of complex 3D features and increased prediction of object labels at both low and high resolutions. Our auxiliary learning approach predicts object classes based on the subvolume parts of a 3D object that improve the classification accuracy compared to other existing 3D volumetric CNN methods. In addition, the influence of views and depths of the 3D model on the classification performance is investigated through extensive experiments applied to the ModelNet40 database. Our deep learning framework runs significantly faster and consumes less memory than full voxel representations and demonstrate the effectiveness of our octree-based auxiliary learning approach for exploring high resolution 3D models. Experimental results reveal the superiority of our MV-DNN that achieves better classification accuracy compared to state-of-art methods on two public databases. |
| first_indexed | 2024-12-17T00:46:47Z |
| format | Article |
| id | doaj.art-f2144e0490a8457cba4994fe80ae1c41 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-17T00:46:47Z |
| publishDate | 2020-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-f2144e0490a8457cba4994fe80ae1c412022-12-21T22:09:54ZengIEEEIEEE Access2169-35362020-01-018238022381610.1109/ACCESS.2020.296850689643823D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning ApproachA. A. M. Muzahid0https://orcid.org/0000-0003-2001-1922Wanggen Wan1https://orcid.org/0000-0002-5065-9650Ferdous Sohel2https://orcid.org/0000-0003-1557-4907Naimat Ullah Khan3https://orcid.org/0000-0001-5224-1381Ofelia Delfina Cervantes Villagomez4https://orcid.org/0000-0002-5180-4506Hidayat Ullah5https://orcid.org/0000-0003-0960-4639School of Communications and Information Engineering, Shanghai University, Shanghai, ChinaSchool of Communications and Information Engineering, Shanghai University, Shanghai, ChinaCollege of Science, Health, Engineering and Education, Murdoch University, Murdoch, WA, AustraliaSchool of Communications and Information Engineering, Shanghai University, Shanghai, ChinaComputing, Electronics and Mechatronics Department, Universidad de las Américas Puebla, Puebla, MexicoSchool of Communications and Information Engineering, Shanghai University, Shanghai, ChinaWe consider the recent challenges of 3D shape analysis based on a volumetric CNN that requires a huge computational power. This high-cost approach forces to reduce the volume resolutions when applying 3D CNN on volumetric data. In this context, we propose a multiorientation volumetric deep neural network (MV-DNN) for 3D object classification with octree generating low-cost volumetric features. In comparison to conventional octree representations, we propose to limit the octree partition to a certain depth to reserve all leaf octants with sparsity features. This allows for improved learning of complex 3D features and increased prediction of object labels at both low and high resolutions. Our auxiliary learning approach predicts object classes based on the subvolume parts of a 3D object that improve the classification accuracy compared to other existing 3D volumetric CNN methods. In addition, the influence of views and depths of the 3D model on the classification performance is investigated through extensive experiments applied to the ModelNet40 database. Our deep learning framework runs significantly faster and consumes less memory than full voxel representations and demonstrate the effectiveness of our octree-based auxiliary learning approach for exploring high resolution 3D models. Experimental results reveal the superiority of our MV-DNN that achieves better classification accuracy compared to state-of-art methods on two public databases.https://ieeexplore.ieee.org/document/8964382/3D shape analysisobject classificationconvolutional neural networkDNNsvolumetric CNN |
| spellingShingle | A. A. M. Muzahid Wanggen Wan Ferdous Sohel Naimat Ullah Khan Ofelia Delfina Cervantes Villagomez Hidayat Ullah 3D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning Approach IEEE Access 3D shape analysis object classification convolutional neural network DNNs volumetric CNN |
| title | 3D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning Approach |
| title_full | 3D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning Approach |
| title_fullStr | 3D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning Approach |
| title_full_unstemmed | 3D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning Approach |
| title_short | 3D Object Classification Using a Volumetric Deep Neural Network: An Efficient Octree Guided Auxiliary Learning Approach |
| title_sort | 3d object classification using a volumetric deep neural network an efficient octree guided auxiliary learning approach |
| topic | 3D shape analysis object classification convolutional neural network DNNs volumetric CNN |
| url | https://ieeexplore.ieee.org/document/8964382/ |
| work_keys_str_mv | AT aammuzahid 3dobjectclassificationusingavolumetricdeepneuralnetworkanefficientoctreeguidedauxiliarylearningapproach AT wanggenwan 3dobjectclassificationusingavolumetricdeepneuralnetworkanefficientoctreeguidedauxiliarylearningapproach AT ferdoussohel 3dobjectclassificationusingavolumetricdeepneuralnetworkanefficientoctreeguidedauxiliarylearningapproach AT naimatullahkhan 3dobjectclassificationusingavolumetricdeepneuralnetworkanefficientoctreeguidedauxiliarylearningapproach AT ofeliadelfinacervantesvillagomez 3dobjectclassificationusingavolumetricdeepneuralnetworkanefficientoctreeguidedauxiliarylearningapproach AT hidayatullah 3dobjectclassificationusingavolumetricdeepneuralnetworkanefficientoctreeguidedauxiliarylearningapproach |