상세정보
학술대회
Monocular Depth Estimation for Mobile Device
- 발행일
- 202111
- 출처
- International Conference on Consumer Electronics (ICCE) 2021 : Asia, pp.498-500
- 협약과제
- 21HH5100, 객체추출 및 실-가상 정합 지원 모바일 AR 기술 개발, 고종국
- 초록
- Depth estimation is an important work in inferring scene geometry from 2D images applied in robotics, AR, autonomous driving, etc. Especially, monocular depth estimation is required for wide application in mobile devices. In this paper, a lightweight monocular depth estimation method is proposed for mobile devices and applied to mobile devices by changing the loss function and using network slimming. The result shows that our model requires low memory and reduces computational complexity that is suitable for mobile devices. The mobile device that we deploy is iPhone 11 Pro and it operates at over 30fps.
- KSP 제안 키워드
- Computational complexity, Estimation method, Mobile devices, Monocular depth estimation, autonomous driving, loss function