READ ME File For 'Dataset for Inverse Design of Structural Color: Finding Multiple Solutions via Conditional Generative Adversarial Networks'
Dataset DOI: https://doi.org/10.5258/SOTON/D2182
ReadMe Author: Peng Dai, University of Southampton [0000-0002-5973-9155]
This dataset supports the publication: Inverse Design of Structural Color: Finding Multiple Solutions via Conditional Generative Adversarial Networks
AUTHORS: Peng Dai, Kai Sun, Xingzhao Yan, Otto Muskens, Kees de Groot, Xupeng Zhu, Yueqiang Hu, Huigao Duan, and Ruomeng Huang
TITLE: Inverse Design of Structural Color: Finding Multiple Solutions via Conditional Generative Adversarial Networks
JOURNAL: Nanophotonics
PAPER DOI IF KNOWN: https://doi.org/10.1515/nanoph-2022-0095
This dataset contains:
The raw data of Figure 3, 4, 5, 6 and 7.
The dataset training and test neural network.
The original image of optical photography.
The original images of SEM images.
The figures are as follows:
Figure 3:
The dataset distribution and cGAN training loss curves.
(a–c) The thickness histograms of the entire dataset. (d) The training loss curves of the generator (blue) and evaluator (red). (e) The curves of evaluator loss components including the real (red) and fake (blue) scores. (f) The MSE between Lab predicted by generator and ground truth (red), and generator’s fake score (blue).
Figure 4:
The distribution comparison between ground truth and prediction. The thickness histograms of (a–c) the testing set and (d–f) predicted by the generator.
Figure 5:
The tendencies of solution group number and ΔE as z sampling number varies.
(a and b) The solution group number histograms when each Lab combines 100 (a) and 1000 z (b). (c) The curve of average testing solution group number against z number, the circle and triangle markers refer to the solution group numbers when each Lab assigned with 1000 and 2100 z, respectively. (d and e) The testing ΔE histograms when each Lab is assigned with 1 (d) and 1000 (e) z, respectively. (f) The curve of average testing ΔE against z number, in which the insert is the enlarge of yellow shading region, the circle and triangle markers refer to the ΔE while each Lab assigned with the z number of 15 and 400.
Figure 6:
The analysis of sRGB color filter design results.
(a–c) The MSE curves of the colors with sRGB values of (a) (0.5, 0, 0 red), (b) (0, 0.5, 0 green), and (c) (0, 0, 0.5 blue) when the SiO2 thickness (d 2) was swept from 0 to 1000 nm and Ag thicknesses (d 1 and d 3) were fixed at 30 nm. The DBSCAN clustered predicted d 2 histograms for the (d) red, (e) green and (f) blue colors. The dark-colored bar indicates the lower resonant order, and the light-colored bar means the higher order.
Figure 7:
The experimental results of fabricated sRGB color filters.
(a–c) The cross-sectional SEM images of the fabricated color filters with all the scale bars of 100 nm. (d–f) The measured spectra (solid line) and corresponding theoretical ones (dash line). (g–i) The color reconstructions, wherein TGT. is the target color; DSG. is the theoretical color of the designed color filter; EXPT. SPT. is the theoretical color of the measured spectrum and EXPT. PHT. is the photograph of the sample taken by the camera.
Date of data collection: November 2020 to September 2021
Information about geographic location of data collection: United Kingdom, China
Licence: CC-BY
Related projects: International Exchange Scheme (IEC\NSFC\170193) between Royal Society (UK) and the National Natural Science Foundation of China (China); This work is a part of the ADEPT project funded by a program grant from the EPSRC (EP/N035437/1).
Date that the file was created: September, 2022