READ ME File For 'Dataset for In-flight sensing of pollen grains via laser scattering and deep learning'

Dataset DOI: 10.5258/SOTON/D1667

ReadMe Author: James A Grant-Jacob  University of Southampton https://orcid.org/0000-0002-4270-4247

This dataset supports the publication:
AUTHORS:James A. Grant-Jacob, Matthew Praeger, Robert W. Eason and Ben Mills
TITLE:In-flight sensing of pollen grains via laser scattering and deep learning

JOURNAL: IOP Engineering Research Express
PAPER DOI IF KNOWN: 


This dataset contains:

Fig1.png
Fig2.png
Fig3.png
Fig4.png
Fig5.png
Fig6.png

Fig6a.txt Data for the graph in figure 6a of the pollen species/class prediction of each frame.

The figures are as follows:

Fig1. Schematics of the experimental setup used for (a) training and (b) testing the neural networks, showing the illumination of pollen grains using three different lasers and the subsequent recording of the scattering pattern by a camera. For testing, the glass slide was removed and pollen grains were dispersed through the beam via the use of a funnel to allow the pollen grains to fall through the laser beam waist. Left inset: experimentally recorded image and corresponding scattering pattern of Narcissus pollen grain captured in the experimental setup. Right inset: in-flight scattering pattern during Narcissus pollen grain dispersing.

Fig2. Images of (a) 10 Populus deltoides and (b) 10 Narcissus pollen grains from which laser light was scattered for training of the neural networks.

Fig3. Concept of determining a pollen species and generating images of pollen grains from their scattering pattern using neural networks. The CNN was trained to transform a scattering pattern into a species label. The cGAN was trained to transform a scattering pattern into a reconstructed image of the pollen grain, along with identification of its XYZ positions. 

Fig4. a) Confusion matrix demonstrating the ability of a neural network to predict the pollen species from previously unseen scattering patterns of Populus deltoides pollen grains, Narcissus pollen grains and the null category. Scattering pattern with the neural network, experimental microscope image, and image generated by the cGAN for (b) Populus deltoides and (c) Narcissus pollen grains and (d) the null category. The colour bars and their RGB value at the left of the experimental and generated images relate to the XYZ position of the pollen grain in 3D space, respectively.

Fig5. (a) Cropped scattering pattern with no pollen grain present (null category), (b) cropped scattering pattern from a Narcissus pollen grain and (c) corresponding activation map from the cropped scattering pattern in (b).

Fig6. Capability of the trained neural network to identify pollen grains in-flight for the Populus deltoides, Narcissus and null pollen grains, showing the prediction for each of the 200 frames (scattering pattern images) in which Populus deltoides (1-100 frames) and Narcissus (101-200 frames) pollen grains were dispersed over the laser beam focus. Included is the absolute change in summed frame intensity (normalised to 1), indicating the likely presence of pollen grain(s) within the beam. (b) Generated images from in-flight captured scattering patterns, showing Populus deltoides, Narcissus and null pollen. 

Date of data collection: 10/10/2020

Information about geographic location of data collection: Building 46, University of Southampton, Southampton, SO17 1BJ
 
Licence: CC-BY

Related projects:
EPSRC grant EP/N03368X/1
EPSRC grant EP/T026197/1


Date that the file was created: 04, 2021