Nanophotonics · 2025 · 10 citations · 38 references
ComputingComputer ScienceMaterials ScienceComputer VisionMachine VisionVisual PerceptionImagingObject DetectionObject RecognitionDevice MiniaturizationLensesMiniaturizationObject Classification
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 cm 3 , 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.
38
You Only Look Once: Unified, Real-Time Object Detection
Joseph Redmon, Santosh Divvala, Ross Girshick et al. · 2016 · 3.3K citations · Full text
Polarization-Controlled Tunable Directional Coupling of Surface Plasmon Polaritons
Jiao Lin, J. P. Balthasar Mueller, Qian Wang et al. · Science · 2013 · 1.2K citations