KSC 2019
Current Result Document : 15 / 15
| 한글제목(Korean Title) |
다중 작업 학습을 통한 이벤트에서 HDR, 심도 및 동영상 이미지 추출 |
| 영문제목(English Title) |
Event to HDR, Depth and Motion Images by Multi Task Learning |
| 저자(Author) |
S. Mohammad Mostfavi
최종현
윤국진
Mostafavi Mohammad
JongHyun Choi
Kuk-Jin Yoon
|
| 원문수록처(Citation) |
VOL 46 NO. 02 PP. 0473 ~ 0475 (2019. 12) |
한글내용
(Korean Abstract) |
|
영문내용
(English Abstract) |
Event cameras are the new capturing device that offers high dynamic range, low latency and low power consumption. Since these cameras do not directly report an intensity view of the scene a new field of research has merged with the goal of synthesizing image-like outputs directly from the event stream. We explore the mutual usefulness of estimating more than one output, namely the intensity image and extend it to further learning depth and optical flow under one network. Through our experiments we show that the simultaneous multitasking of the learning process provides higher quality intensity images in terms of the PSNR which were not been able to reach through learning a signle task.
|
| 키워드(Keyword) |
|
| 파일첨부 |
PDF 다운로드
|
