EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGES
This paper investigates the value of depth modality in object classification in RGB-D images. We use a simple model based on a multi-layered convolutional neural network which we train on a dataset of segmented RGB-D images of household and office objects. We evaluate and quantify the benefit of add...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Sciendo
2019-01-01
|
| Series: | Acta Electrotechnica et Informatica |
| Subjects: | |
| Online Access: | http://www.aei.tuke.sk/papers/2018/4/04_Varga.pdf |
| _version_ | 1797711874313158656 |
|---|---|
| author | Michal VARGA Ján JADLOVSKÝ |
| author_facet | Michal VARGA Ján JADLOVSKÝ |
| author_sort | Michal VARGA |
| collection | DOAJ |
| description | This paper investigates the value of depth modality in object classification in RGB-D images. We use a simple model based on a multi-layered convolutional neural network which we train on a dataset of segmented RGB-D images of household and office objects. We evaluate and quantify the benefit of additional depth modality and its effect on classification accuracy on this dataset. Also, we compare the benefit of depth channel against the addition of color to grayscale image. Our experimental results support a conclusion, that for these categories of objects the depth modality provides a significant benefit to classification, which also outweighs the benefit of color information. Similar supporting evidence found in recent research is shown in comparison along with the resulting quantified benefit of depth modality. |
| first_indexed | 2024-03-12T07:13:30Z |
| format | Article |
| id | doaj.art-a6f2018459a34af6b12d787059048657 |
| institution | Directory Open Access Journal |
| issn | 1335-8243 1338-3957 |
| language | English |
| last_indexed | 2024-03-12T07:13:30Z |
| publishDate | 2019-01-01 |
| publisher | Sciendo |
| record_format | Article |
| series | Acta Electrotechnica et Informatica |
| spelling | doaj.art-a6f2018459a34af6b12d7870590486572023-09-02T22:57:19ZengSciendoActa Electrotechnica et Informatica1335-82431338-39572019-01-01184263110.15546/aeei-2018-0029EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGESMichal VARGA0 Ján JADLOVSKÝ1Department of Cybernetics and Artificial Intelligence, Faculty of Electrical Engineering and Informatics, Technical University of Kosice, Letna 9, 042 00 Kosice, Slovak RepublicDepartment of Cybernetics and Artificial Intelligence, Faculty of Electrical Engineering and Informatics, Technical University of Kosice, Letna 9, 042 00 Kosice, Slovak RepublicThis paper investigates the value of depth modality in object classification in RGB-D images. We use a simple model based on a multi-layered convolutional neural network which we train on a dataset of segmented RGB-D images of household and office objects. We evaluate and quantify the benefit of additional depth modality and its effect on classification accuracy on this dataset. Also, we compare the benefit of depth channel against the addition of color to grayscale image. Our experimental results support a conclusion, that for these categories of objects the depth modality provides a significant benefit to classification, which also outweighs the benefit of color information. Similar supporting evidence found in recent research is shown in comparison along with the resulting quantified benefit of depth modality.http://www.aei.tuke.sk/papers/2018/4/04_Varga.pdf3D imagingcomputer visionconvolutional neural networkdeep learning |
| spellingShingle | Michal VARGA Ján JADLOVSKÝ EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGES Acta Electrotechnica et Informatica 3D imaging computer vision convolutional neural network deep learning |
| title | EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGES |
| title_full | EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGES |
| title_fullStr | EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGES |
| title_full_unstemmed | EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGES |
| title_short | EVALUATION OF DEPTH MODALITY IN CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION OF RGB-D IMAGES |
| title_sort | evaluation of depth modality in convolutional neural network classification of rgb d images |
| topic | 3D imaging computer vision convolutional neural network deep learning |
| url | http://www.aei.tuke.sk/papers/2018/4/04_Varga.pdf |
| work_keys_str_mv | AT michalvarga evaluationofdepthmodalityinconvolutionalneuralnetworkclassificationofrgbdimages AT janjadlovsky evaluationofdepthmodalityinconvolutionalneuralnetworkclassificationofrgbdimages |