Collision of high-resolution wide FOV metalens cameras and vision tasks

Abstract

Metalenses, with their compact form factor and unique optical capabilities, hold tremendous potential for advancing computer vision applications. In this work, we propose a high-resolution, large field-of-view (FOV) metalens intelligent recognition system, combining the latest YOLO framework, aimed at supporting a range of vision tasks. Specifically, we demonstrate its effectiveness in scanning, pose recognition, and object classification. The metalens we designed to achieve a 100° FOV while operating near the diffraction limit, as confirmed by experimental results. Moreover, the metalenses weigh only 0.1 g and occupy a compact volume of 0.04 cm3, effectively addressing the bulkiness of conventional lenses and overcoming the limitations of traditional metalens in spatial frequency transmission. This work highlights the transformative potential of metalenses in the field of computer vision, The integration of metalenses with computer vision opens exciting possibilities for next-generation imaging systems, offering both enhanced functionality and unprecedented miniaturization.