Presentation | 2015-03-19 A Study on Efficient Pedestrian Detection by Combining CNN and SVM Takuma SAITO, Yuji WAIZUMI, Kazuyuki TANAKA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Detecting pedestrian in real time is challenging problem for application to road safety. For high-accuracy and high-speed pedestrian detection, we propose a efficient pedestrian detection system by combining Convolutional Neural Network (CNN) and Support Vector Machine (SVM). To compress and select the features, we use CNN which can extract compressed features from input space by its convolution operation. The compressed features include information which separates pedestrian from background image effectively due to the supervised learning nature of CNN. By sorting them out, we can obtain more compact features for pedestrian detection. Linear SVM and nonlinear SVM are used to build a hierarchical detection system. For high-speed detection in nonlinear SVM phase, the compressed features are inputted to nonlinear SVM. In detection experiment, we demonstrated that our proposed method can achieve higher detection accuracy than linear SVM and shorter computation time than nonlinear SVM. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | pedestrian detection / CNN / SVM / feature selection |
Paper # | BioX2014-46,PRMU2014-166 |
Date of Issue |
Conference Information | |
Committee | PRMU |
---|---|
Conference Date | 2015/3/12(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Pattern Recognition and Media Understanding (PRMU) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study on Efficient Pedestrian Detection by Combining CNN and SVM |
Sub Title (in English) | |
Keyword(1) | pedestrian detection |
Keyword(2) | CNN |
Keyword(3) | SVM |
Keyword(4) | feature selection |
1st Author's Name | Takuma SAITO |
1st Author's Affiliation | Graduate School of Information Science, Tohoku University() |
2nd Author's Name | Yuji WAIZUMI |
2nd Author's Affiliation | Graduate School of Information Science, Tohoku University |
3rd Author's Name | Kazuyuki TANAKA |
3rd Author's Affiliation | Graduate School of Information Science, Tohoku University |
Date | 2015-03-19 |
Paper # | BioX2014-46,PRMU2014-166 |
Volume (vol) | vol.114 |
Number (no) | 521 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |