| Presentation | 2018-10-25 Improvement of Classification Accuracy for Imbalanced Training Data by CasNet Takuro Oki, Ryusuke Miyamoto, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Imbalanced samples composed of limited number of positive samples corresponding to objects and huge number of negative samples extracted from background regions reduces the accuracy of visual object detection. To solve this problem, this paper proposes a novel convolutional neural network named "CasNet". CasNet introduces cascade structure that is used for rapid and accurate object detector in order to reduce the number of negative. The CasNet become a cascade stage when it is attached to a layer of existing convolutinoal neural networks to construct cascaded classifier. Each stage composed of a CasNet peforms two-class classification to reject easy negatives corresponding to background regions. By this early rejection of easy negatives, a main network can be trained to classify more complex samples. Experimental results using a dataset created from the PASCAL VOC2012 dataset showed that higher accuracy was obtained at less training iterations if CasNets were attached to VGG16 appropriately. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Visual object detection / Convolutional neural network / Data imbalance problem |
| Paper # | SIS2018-13 |
| Date of Issue | 2018-10-18 (SIS) |
| Conference Information | |
| Committee | SIS / ITE-BCT |
|---|---|
| Conference Date | 2018/10/25(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Kyoto University Clock Tower Centennial Hall |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | System Implementation Technology, Short Range Wireless Systems, Smart Multimedia Systems, Broadcasting Technology, etc. |
| Chair | Takayuki Nakachi(NTT) / Tomoaki Otsuki(Keio Univ) |
| Vice Chair | Noriaki Suetake(Yamaguchi Univ.) / Tomoaki Kimura(Kanagawa Inst. of Tech.) / Kyoichi Saito(NHK) / Yasushi Kasuga(TV Asahi) |
| Secretary | Noriaki Suetake(Kyushu Inst. of Tech.) / Tomoaki Kimura(Tokyo Metropolitan Univ.) / Kyoichi Saito(B-SAT) / Yasushi Kasuga(NHK) |
| Assistant | Takanori Koga(National Inst. of Tech. Tokuyama College) / Hideaki Misawa(National Inst. of Tech., Ube College) / Shigeki Shiokawa(Kanagawa Inst. of Tech.) / Toshiharu Morizumi(NTT) / Iwao Namikawa(Kansai Telecasting Corporation) |
| Paper Information | |
| Registration To | Technical Committee on Smart Info-Media Systems / Technical Group on Broadcasting Technology |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Improvement of Classification Accuracy for Imbalanced Training Data by CasNet |
| Sub Title (in English) | |
| Keyword(1) | Visual object detection |
| Keyword(2) | Convolutional neural network |
| Keyword(3) | Data imbalance problem |
| 1st Author's Name | Takuro Oki |
| 1st Author's Affiliation | Meiji University(Meiji Univ.) |
| 2nd Author's Name | Ryusuke Miyamoto |
| 2nd Author's Affiliation | Meiji University(Meiji Univ.) |
| Date | 2018-10-25 |
| Paper # | SIS2018-13 |
| Volume (vol) | vol.118 |
| Number (no) | SIS-264 |
| Page | pp.pp.19-24(SIS), |
| #Pages | 6 |
| Date of Issue | 2018-10-18 (SIS) |