Fig. 5: Histograms (1 nm2 ∙ bin−1) and KDE-plots of calculated nanoparticle areas (nm2) for the ground truth and inferred datasets for the Au/Ge system.
From: nNPipe: a neural network pipeline for automated analysis of morphologically diverse catalyst systems
KDE-plots were generated using a Gaussian Kernel and Scott’s rule53. Tables containing the calculated mean nanoparticle area (nm2), nanoparticle counts in the dataset (|D|) and calculated errors with respect to the ground truth are also shown. A minimum nanoparticle size of 50 pixels was assumed for all images. a The full dataset, b Images with an F1-Score of >0.9, c Images with an F1-Score of >0.8, d Images with an F1-Score of >0.7 and e Images with an F1-Score of >0.6.
