| Presentation | 2019-01-18 GANs for Generating Whole Image from One Region of 360-degree Naofumi Akimoto, Masaki Hayashi, Seito Kasai, Yoshimitsu Aoki, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper, we present a novel problem setting in which, using one direction of a 360-degree image, a Generative Adversarial Networks (GANs) completes a whole 360-degree image. We also address this problem with a goal which is, distortions of the outlines of roads and buildings that specifically exist in 360-degree images should be generated. Furthermore, for making this problem easy, we present image rearranging which is done using a specific property seen in a 360-degree image. This is that both edge of 360-degree images are originally continuous. And also, we present a combination of dilated convolution layers as effective architectures for generation of a 360-degree image. In our experiments, we show that the series and/or parallel architecture generated better results, in which the white holes seen in baseline results were suppressed and the distortions of the outlines of roads and buildings were generated. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Generative Adversarial Network / deep learning |
| Paper # | PRMU2018-111,MVE2018-53 |
| Date of Issue | 2019-01-10 (PRMU, MVE) |
| Conference Information | |
| Committee | PRMU / MVE / IPSJ-CVIM |
|---|---|
| Conference Date | 2019/1/17(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Shinichi Sato(NII) / Kenji Mase(Nagoya Univ.) |
| Vice Chair | Yoshihisa Ijiri(Omron) / Toru Tamaki(Hiroshima Univ.) / Masayuki Ihara(NTT) |
| Secretary | Yoshihisa Ijiri(NEC) / Toru Tamaki(Osaka Univ.) / Masayuki Ihara(NTT) / (Kyushu Univ.) |
| Assistant | Go Irie(NTT) / Yoshitaka Ushiku(Univ. of Tokyo) / Satoshi Nishiguchi(Oosaka Inst. of Tech.) / Masanori Yokoyama(*) |
| Paper Information | |
| Registration To | Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Media Experience and Virtual Environment / Special Interest Group on Computer Vision and Image Media |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | GANs for Generating Whole Image from One Region of 360-degree |
| Sub Title (in English) | |
| Keyword(1) | Generative Adversarial Network |
| Keyword(2) | deep learning |
| Keyword(3) | |
| 1st Author's Name | Naofumi Akimoto |
| 1st Author's Affiliation | Keio University(Keio Univ.) |
| 2nd Author's Name | Masaki Hayashi |
| 2nd Author's Affiliation | Keio University(Keio Univ.) |
| 3rd Author's Name | Seito Kasai |
| 3rd Author's Affiliation | Keio University(Keio Univ.) |
| 4th Author's Name | Yoshimitsu Aoki |
| 4th Author's Affiliation | Keio University(Keio Univ.) |
| Date | 2019-01-18 |
| Paper # | PRMU2018-111,MVE2018-53 |
| Volume (vol) | vol.118 |
| Number (no) | PRMU-404,MVE-405 |
| Page | pp.pp.149-153(PRMU), pp.149-153(MVE), |
| #Pages | 5 |
| Date of Issue | 2019-01-10 (PRMU, MVE) |