Localized deep extreme learning machines for efficient RGB-D object recognition
Existing RGB-D object recognition methods either use channel specific handcrafted features, or learn features with deep networks. The former lack representation ability while the latter require large amounts of training data and learning time. In real-time robotics applications involving RGB-D senso...
| Main Authors: | , , |
|---|---|
| Format: | Proceeding Paper |
| Language: | English English |
| Published: |
Institute of Electrical and Electronic Engineers, Inc. (IEEE)
2015
|
| Subjects: | |
| Online Access: | http://irep.iium.edu.my/64704/ http://irep.iium.edu.my/64704/7/64704%20Localized%20Deep%20Extreme%20Learning.pdf http://irep.iium.edu.my/64704/8/64704%20Localized%20Deep%20Extreme%20Learning%20SCOPUS.pdf |
| _version_ | 1848786262046015488 |
|---|---|
| author | Mohd Zaki, Hasan Firdaus Shafait, Faisal Mian, Ajmal S. |
| author_facet | Mohd Zaki, Hasan Firdaus Shafait, Faisal Mian, Ajmal S. |
| author_sort | Mohd Zaki, Hasan Firdaus |
| building | IIUM Repository |
| collection | Online Access |
| description | Existing RGB-D object recognition methods either use channel specific handcrafted features, or learn features with deep networks. The former lack representation ability while the latter require large amounts of training data and learning time. In real-time robotics applications involving RGB-D sensors, we do not have the luxury of both. In this paper, we propose Localized Deep Extreme Learning Machines (LDELM) that efficiently learn features from RGB-D data. By using localized patches, not only is the problem of data sparsity solved, but the learned features are robust to occlusions and viewpoint variations. LDELM learns deep localized features in an unsupervised way from random patches of the training data. Each image is then feed-forwarded, patch-wise, through the LDELM to form a cuboid of features. The cuboid is divided into cells and pooled to get the final compact image representation which is then used to train an ELM classifier. Experiments on the benchmark Washington RGB-D and 2D3D datasets show that the proposed algorithm not only is significantly faster to train but also outperforms state-of-the-art methods in terms of accuracy and classification time. |
| first_indexed | 2025-11-14T17:06:14Z |
| format | Proceeding Paper |
| id | iium-64704 |
| institution | International Islamic University Malaysia |
| institution_category | Local University |
| language | English English |
| last_indexed | 2025-11-14T17:06:14Z |
| publishDate | 2015 |
| publisher | Institute of Electrical and Electronic Engineers, Inc. (IEEE) |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | iium-647042018-08-06T07:02:51Z http://irep.iium.edu.my/64704/ Localized deep extreme learning machines for efficient RGB-D object recognition Mohd Zaki, Hasan Firdaus Shafait, Faisal Mian, Ajmal S. QA75 Electronic computers. Computer science Existing RGB-D object recognition methods either use channel specific handcrafted features, or learn features with deep networks. The former lack representation ability while the latter require large amounts of training data and learning time. In real-time robotics applications involving RGB-D sensors, we do not have the luxury of both. In this paper, we propose Localized Deep Extreme Learning Machines (LDELM) that efficiently learn features from RGB-D data. By using localized patches, not only is the problem of data sparsity solved, but the learned features are robust to occlusions and viewpoint variations. LDELM learns deep localized features in an unsupervised way from random patches of the training data. Each image is then feed-forwarded, patch-wise, through the LDELM to form a cuboid of features. The cuboid is divided into cells and pooled to get the final compact image representation which is then used to train an ELM classifier. Experiments on the benchmark Washington RGB-D and 2D3D datasets show that the proposed algorithm not only is significantly faster to train but also outperforms state-of-the-art methods in terms of accuracy and classification time. Institute of Electrical and Electronic Engineers, Inc. (IEEE) 2015 Proceeding Paper PeerReviewed application/pdf en http://irep.iium.edu.my/64704/7/64704%20Localized%20Deep%20Extreme%20Learning.pdf application/pdf en http://irep.iium.edu.my/64704/8/64704%20Localized%20Deep%20Extreme%20Learning%20SCOPUS.pdf Mohd Zaki, Hasan Firdaus and Shafait, Faisal and Mian, Ajmal S. (2015) Localized deep extreme learning machines for efficient RGB-D object recognition. In: 2015 International Conference on Digital Image Computing: Techniques and Applications (DICTA), 23rd-25th November 2015, Adelaide, SA, Australia. https://ieeexplore.ieee.org/document/7371280/ 10.1109/DICTA.2015.7371280 |
| spellingShingle | QA75 Electronic computers. Computer science Mohd Zaki, Hasan Firdaus Shafait, Faisal Mian, Ajmal S. Localized deep extreme learning machines for efficient RGB-D object recognition |
| title | Localized deep extreme learning machines for efficient RGB-D object recognition |
| title_full | Localized deep extreme learning machines for efficient RGB-D object recognition |
| title_fullStr | Localized deep extreme learning machines for efficient RGB-D object recognition |
| title_full_unstemmed | Localized deep extreme learning machines for efficient RGB-D object recognition |
| title_short | Localized deep extreme learning machines for efficient RGB-D object recognition |
| title_sort | localized deep extreme learning machines for efficient rgb-d object recognition |
| topic | QA75 Electronic computers. Computer science |
| url | http://irep.iium.edu.my/64704/ http://irep.iium.edu.my/64704/ http://irep.iium.edu.my/64704/ http://irep.iium.edu.my/64704/7/64704%20Localized%20Deep%20Extreme%20Learning.pdf http://irep.iium.edu.my/64704/8/64704%20Localized%20Deep%20Extreme%20Learning%20SCOPUS.pdf |