| Presentation | 2022-03-29 Weighted Dice Loss for Segmentation from Noisy Labels Toshikazu Samura, Katsumi Tadamura, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Deep neural networks (DNNs) are easily affected by noisy labels during training. It degrades the classification performance of DNNs. In this study, we proposed weighted Dice loss for decreasing the adverse effects of noisy labels in a segmentation task. The cleanness of label is reflected to the learning progress. The weighted Dice loss reduces the weights for presumed noisy pixels according to the current epoch and their learning progress. We applied the U-Net trained through the proposed loss into road extraction from aerial images. Consequently, we demonstrated that the weighted Dice loss decreases the adverse effects of noisy labels in the segmentation task. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Weighted Dice loss / Noisy label / U-Net / Segmentation / Road extraction |
| Paper # | MSS2021-77,NLP2021-148 |
| Date of Issue | 2022-03-21 (MSS, NLP) |
| Conference Information | |
| Committee | MSS / NLP |
|---|---|
| Conference Date | 2022/3/28(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Online |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | MSS, NLP, Work In Progress (MSS only), and etc. |
| Chair | Atsuo Ozaki(Osaka Inst. of Tech.) / Takuji Kosaka(Chukyo Univ.) |
| Vice Chair | Shingo Yamaguchi(Yamaguchi Univ.) / Akio Tsuneda(Kumamoto Univ.) |
| Secretary | Shingo Yamaguchi(Hokkaido Univ.) / Akio Tsuneda(NEC) |
| Assistant | Masato Shirai(Shimane Univ.) / Hideyuki Kato(Oita Univ.) / Yuichi Yokoi(Nagasaki Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Mathematical Systems Science and its Applications / Technical Committee on Nonlinear Problems |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Weighted Dice Loss for Segmentation from Noisy Labels |
| Sub Title (in English) | |
| Keyword(1) | Weighted Dice loss |
| Keyword(2) | Noisy label |
| Keyword(3) | U-Net |
| Keyword(4) | Segmentation |
| Keyword(5) | Road extraction |
| 1st Author's Name | Toshikazu Samura |
| 1st Author's Affiliation | Yamaguchi University(Yamaguchi Univ.) |
| 2nd Author's Name | Katsumi Tadamura |
| 2nd Author's Affiliation | Yamaguchi University(Yamaguchi Univ.) |
| Date | 2022-03-29 |
| Paper # | MSS2021-77,NLP2021-148 |
| Volume (vol) | vol.121 |
| Number (no) | MSS-443,NLP-444 |
| Page | pp.pp.117-120(MSS), pp.117-120(NLP), |
| #Pages | 4 |
| Date of Issue | 2022-03-21 (MSS, NLP) |