Particle Sensing via Deep Learning
Particle Sensing via Deep Learning
Airborne pollution, in the form of particles of sizes <10 μm (PM10) and <2.5 μm (PM2.5), is a global health challenge and linked to illnesses such as asthma, cancer, heart disease and dementia, contributing to around nine million deaths globally per year. Effective monitoring of airborne particles, particularly in residential areas, is urgently needed to better understand personal exposure and pollution hotspots. Whilst existing in-the-field sensors can provide an approximate count of the number of airborne particles, they are unable to identify properties such as the material type and shape of individual particles. The capability for parameter identification of individual particles is essential, as different pollution particle parameters are associated with specific health conditions. The interface of deep learning and imaging has seen extraordinary progress in the past few years, as computational power now enables image processing that can exceed human capability. The hypothesis is that deep learning will enable the creation of a low-cost, compact, real-time sensing device capable of simultaneous identification and imaging of individual airborne pollution particles. The focus of this talk will be on the application of convolutional neural networks for the identification of material type, shape and size of individual pollution particles directly from scattering patterns, as shown by the concept figure. Results for real-time sensing for both airborne and waterborne particles, and using both free-space lasers (including powered by a Raspberry Pi) and fibre optic bundles will be presented.
Mills, Benjamin
05f1886e-96ef-420f-b856-4115f4ab36d0
Jain, Saurabh
bf4af598-26bf-47f4-a0a4-800095a23eb5
Xie, Yunhui
f2c3b0e4-8650-4e04-80e5-04505f45bdd6
MacKay, Benita Scout
318d298f-5b38-43d7-b30d-8cd07f69acd4
McDonnell, Michael David Tom
bc7b6423-bd77-424d-81e7-4e5448e926cb
Praeger, Matthew
84575f28-4530-4f89-9355-9c5b6acc6cac
Richardson, David J.
ebfe1ff9-d0c2-4e52-b7ae-c1b13bccdef3
Eason, R.W.
e38684c3-d18c-41b9-a4aa-def67283b020
Grant-Jacob, James
c5d144d8-3c43-4195-8e80-edd96bfda91b
3 December 2020
Mills, Benjamin
05f1886e-96ef-420f-b856-4115f4ab36d0
Jain, Saurabh
bf4af598-26bf-47f4-a0a4-800095a23eb5
Xie, Yunhui
f2c3b0e4-8650-4e04-80e5-04505f45bdd6
MacKay, Benita Scout
318d298f-5b38-43d7-b30d-8cd07f69acd4
McDonnell, Michael David Tom
bc7b6423-bd77-424d-81e7-4e5448e926cb
Praeger, Matthew
84575f28-4530-4f89-9355-9c5b6acc6cac
Richardson, David J.
ebfe1ff9-d0c2-4e52-b7ae-c1b13bccdef3
Eason, R.W.
e38684c3-d18c-41b9-a4aa-def67283b020
Grant-Jacob, James
c5d144d8-3c43-4195-8e80-edd96bfda91b
Mills, Benjamin, Jain, Saurabh, Xie, Yunhui, MacKay, Benita Scout, McDonnell, Michael David Tom, Praeger, Matthew, Richardson, David J., Eason, R.W. and Grant-Jacob, James
(2020)
Particle Sensing via Deep Learning.
78th GASG Colloquium, “Airborne hazards: from gas to particles”.
03 Dec 2020.
Record type:
Conference or Workshop Item
(Other)
Abstract
Airborne pollution, in the form of particles of sizes <10 μm (PM10) and <2.5 μm (PM2.5), is a global health challenge and linked to illnesses such as asthma, cancer, heart disease and dementia, contributing to around nine million deaths globally per year. Effective monitoring of airborne particles, particularly in residential areas, is urgently needed to better understand personal exposure and pollution hotspots. Whilst existing in-the-field sensors can provide an approximate count of the number of airborne particles, they are unable to identify properties such as the material type and shape of individual particles. The capability for parameter identification of individual particles is essential, as different pollution particle parameters are associated with specific health conditions. The interface of deep learning and imaging has seen extraordinary progress in the past few years, as computational power now enables image processing that can exceed human capability. The hypothesis is that deep learning will enable the creation of a low-cost, compact, real-time sensing device capable of simultaneous identification and imaging of individual airborne pollution particles. The focus of this talk will be on the application of convolutional neural networks for the identification of material type, shape and size of individual pollution particles directly from scattering patterns, as shown by the concept figure. Results for real-time sensing for both airborne and waterborne particles, and using both free-space lasers (including powered by a Raspberry Pi) and fibre optic bundles will be presented.
This record has no associated files available for download.
More information
Published date: 3 December 2020
Venue - Dates:
78th GASG Colloquium, “Airborne hazards: from gas to particles”, 2020-12-03 - 2020-12-03
Identifiers
Local EPrints ID: 472027
URI: http://eprints.soton.ac.uk/id/eprint/472027
PURE UUID: a1848f97-f7c0-4594-be0d-e96bf23f45b5
Catalogue record
Date deposited: 24 Nov 2022 17:30
Last modified: 21 Jun 2023 01:52
Export record
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics
