Dataset of: Seeing Beyond RGB Capabilities: Data-Driven and Physics-Guided Broadband Spectral Extrapolation of Plasmonic Nanostructures by Deep Learning

Dataset Description: Seeing Beyond RGB Capabilities: Data-Driven and Physics-Guided Broadband Spectral Extrapolation of Plasmonic Nanostructures by Deep LearningThis dataset supports the development and validation of SPARX, a deep-learning-powered paradigm for the high-throughput characterization of plasmonic nanostructures. It contains: Information-limited RGB images (<700 nm) of various nanoparticles. Broadband dark-field spectra (500–1000 nm) providing ground-truth resonance data. Nanoparticle shape classifications used for batch-processing and morphological analysis. The data demonstrates a data-driven and physics-guided approach to extrapolating broadband spectral responses from limited optical inputs. By learning the physical relationships among multiple orders of resonances, the SPARX framework achieves a speedup of 2–4 orders of magnitude over traditional hyperspectral imaging while maintaining comparable precision. This collection is intended to facilitate research into reproducible nanophotonics and the automation of optical characterization workflows. Funding Sources This work was supported by the National Key Research and Development Program of China (Grant No. 2024YFA1409900) and the National Natural Science Foundation of China (Grant No. 62475071 and 52488301). M.K., and F.P. acknowledge funding by European Union's Horizon Europe under grant agreement 101125498, project MINING - “Multifunctional nano-bio interfaces with deep brain regions”. M.K., and F. P. acknowledge funding from the European Union's Horizon 2020 Research and Innovation Programme project DEEPER under grant agreement 101016787.