| Presentation | 2010-09-22 A Study on Performance Improvement of Generic Object Recognition by Object Extraction Tetsuya FUJIKAWA, Jiro KATTO, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In Bag-of-Keypoints method used in generic object recognition, we ordinarily derive SIFT features over the whole regions of images. However, in some categories, information of background regions is not effective for recognition. Therefore, in this paper, we investigate the effect of object extraction on the recognition rate of Bag-of-Keypoints. Firstly, by using ground truth data of foreground objects, we found that classification rate can be improved by 7% in average and region extraction is effective for some categories but not for other categories. Secondly, we propose a method to extract object regions automatically by using visual attention model prior to Bag-of-Keypoints. Unfortunately, this method shows only 1.5% improvement mainly due to categories which need background regions. Finally, we propose an adaptive method which selects foreground regions or whole image for SIFT feature description according to categories. Experimental results show that above 7% improvement is furthermore expected against other methods described above. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Generic Object Recognition / Bag-of-Keypoints / Object Extraction / Visual Attention Model |
| Paper # | LOIS2010-28,IE2010-70 |
| Date of Issue |
| Conference Information | |
| Committee | LOIS |
|---|---|
| Conference Date | 2010/9/14(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 | Life Intelligence and Office Information Systems (LOIS) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | A Study on Performance Improvement of Generic Object Recognition by Object Extraction |
| Sub Title (in English) | |
| Keyword(1) | Generic Object Recognition |
| Keyword(2) | Bag-of-Keypoints |
| Keyword(3) | Object Extraction |
| Keyword(4) | Visual Attention Model |
| 1st Author's Name | Tetsuya FUJIKAWA |
| 1st Author's Affiliation | Graduate School of Fundamental Science and Engineering, Waseda University() |
| 2nd Author's Name | Jiro KATTO |
| 2nd Author's Affiliation | Graduate School of Fundamental Science and Engineering, Waseda University |
| Date | 2010-09-22 |
| Paper # | LOIS2010-28,IE2010-70 |
| Volume (vol) | vol.110 |
| Number (no) | 207 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |