Dataset for the paper "Selective laser cleaning of microscale particles using deep learning"

This dataset is supported the publication 'Selective Laser Cleaning of Microbeads using Deep Learning' in the Journal :Light: Advanced Manufacturing

The parent folder 'dataset' includes two main folders, one is for figures, another is for numerical data used to plot the figures.

1. Figures
This folder includes all the figures inside the publication, which named in "'figure 1.png', 'figure 2.png', 'figure 3.png', 'figure 4.png','figure 5.png', 'figure 6.png'". The graphics denoted as SM figure 1 to SM figure 4 cited within the supplementary materials of the scholarly publication, are accessible under the file name 'SM figure 1.png', 'SM figure 2.png', 'SM figure 3.png', 'SM figure 4.png'.

2. Numerical data
This folder includes six sub-folders named in "'Fig 2c', 'Fig 5b', 'Fig 6', 'SM Table 1', 'SM Table 2', 'SM Table 3'", the first three are used to plot corresponding figure respectively, while the last three, including their corresponding .csv files, form the tables in the supplementary file.

(a) In 'Fig 2c' folder, there are two NPY files, one named 'num_of_removed_BA.npy' indicates number of removed microbead based on the real before/after laser pulse images; the other named 'num_of_removed_BG.npy' indicates number of removed microbead based on the real before/generated-after laser pulse images. These two files are used to plot the confusion matrix in Figure 2c. Each NPY file shape in (600,), can be load with 'numpy.load(NPY_file)' in python.

(b) In 'Fig 4b' folder, there are two NPY files, one named 'fig4_b_i_iii.npy' is used to draw the subplots (i) and (iii) in Figure 4b, the other named 'fig4_b_ii.npy' is used to draw the subplot (ii) in Figure 4b.

- The shape of 'fig4_b_i_iii.npy' is (68,4), containing 4 columns, which indicate number of microbeads remained in experiment, number of microbeads removed in experiment, number of microbeads remained in simulation, number of microbeads removed in simulation from initial state to after total 67 laser pulses respectively.

- The shape of 'fig4_b_ii.npy' is (8,3), containing 3 columns, which indicate different number of removed microbeads (8 different possible states), frequency of corresponding number of microbeads removal in experiment and simulation respectively.

(c) In 'Fig 5' folder, there are two NPY files, one named 'fig5_a.npy' is used to draw the Figure 5a, the other named 'fig5_b.npy' is used to draw the Figure 5b.

- The shape of 'fig5_a.npy' is (67,3), containing 3 columns, which indicate the XY coordinates of each laser pulse and number of microbeads are removed in experiment.

- The shape of 'fig5_b.npy' is (67,3), containing 3 columns, which indicate the XY coordinates of each laser pulse and number of microbeads are removed in simulation.

(d) In 'SM Table 1', there is one CSV file in the folder, with the same name of the folder.
- The shape of 'fig6_a.npy' is (6,5), containing 3 columns, which indicate the comparison of 4 type of laser cleaning method in 'pulse duration, precision, speed, cleaning scale, energy efficiency' aspects, they are 'Continuous wave laser, Nanosecond pulsed laser, Laser-induced plasma/shockwave, Selective laser cleaning with deep learning'

(e) In 'SM Table 2', there is one CSV file in the folder, with the same name of the folder.
- The shape of 'fig6_a.npy' is (53,4), containing 3 columns, which indicate the number of layer used in generator, image layer, image properties and specification of the layer.

(f) In 'SM Table 3', there is one CSV file in the folder, with the same name of the folder.
- The shape of 'fig6_a.npy' is (13,4), containing 3 columns, which indicate the number of layer used in discriminator, image layer, image properties and specification of the layer.