Human-Inspired Camera: A Novel Camera System for Computer Vision
- Resource Type
- Conference
- Authors
- Kumar, Shubham; Mi, Jonathan; Zhang, Qingyuan; Chang, Benjamin; Le, Hao; Khoshabeh, Ramsin; Nguyen, Truong
- Source
- 2021 18th International SoC Design Conference (ISOCC) SoC Design Conference (ISOCC), 2021 18th International. :29-30 Oct, 2021
- Subject
- Components, Circuits, Devices and Systems
Legged locomotion
Computer vision
Three-dimensional displays
Costs
Computational modeling
Estimation
Focusing
computer vision
depth estimation
human vision
- Language
Computer vision models aim to emulate biological design so that systems can perform meaningful tasks. We believe that the underlying processes of the human visual system hold the keys to further improving the performance of such computer vision solutions. This exploratory paper investigates the swaying motion of human vision when walking to develop a novel camera system. We successfully demonstrate that this design is able to improve performance in computer vision tasks, such as monocular depth estimation.