Metasurface and neural network combo enables compact multi-dimensional infrared imaging

bravo_abad · x · 2026-08-21

Huiming Luo et al. combined an infrared metasurface with deep learning to capture spatial, spectral, and polarization information simultaneously. The hardware uses a silicon nanopillar metasurface to encode data into a grayscale pattern, which a residual-attention neural network then reconstructs. This compact camera covers 1150–1650 nm with 167 spectral channels, offering roughly 2x better spectral and 7x better spatial resolution.

Original post →

More from Research

Research channel →