- > 연구성과 > 학술발표
공지사항
| 논문명 |
적대적 생성 신경망을 이용한 단일 RGB 기반 깊이 이미지 생성 |
| 구분 |
국내발표 |
| 저자 |
푸옹, 호서가, 서가가, 허언민, 한우철, 조경은 |
| 국내/국외 |
|
| 학술회의명 |
2019년도 한국멀티미디어학회 춘계학술발표대회 |
| 개최국가 |
대한민국 |
| 주관기관 |
|
| 게재일 |
2019.05 |
| Depth images are important for many applications such as 3D scene reconstruction. This paper proposes a depth image generation method from a single RGB image. The input of the proposed method is a 2D image captured by a conventional camera, which is generally less expensive. Next, a generative adversarial network is employed to generate a depth image estimation from the input image. Experimental results verify that the proposed method provides high-quality depth images from single 2D images. In addition, the proposed depth image generation method does not require high computational resources. We can use a general PC with a moderate-capacity graphics processing unit for handling the calculations. |
|
