Stern, Nathan (2014) Machine vision detection of crop diseases. [USQ Project]
|
Text
Stern_2014.pdf Download (41MB) | Preview |
Abstract
The production of agricultural crops such as wheat is a multi-billion dollar industry. Each year, diseases such as fungal infections can potentially destroy the entire production in a region if the conditions are right. Traditional mitigation has involved the grower's own judgement as the first option, followed by use of trained professionals for confirmation of diagnosis and advice. Often the onset of infection is rapid, and the professionals may not always be available to assess
the situation before the infection spreads. This reliance on individual judgement and off-site experts highlights a need to develop a reliable, automated software solution which can provide an accurate and immediate software diagnosis at the first sign of infection.
This research has developed into two solutions: a system to help agronomists by using professional camera equipment, as well as the outline for a mobile solution which can operate on a `smart' device to provide growers with an on-hand
diagnosis tool.
By investigating the way the human mind and eye works, a software emulation of the human visual system was constructed, with artificial intelligence approaches used for final interpretation of the optical response. This use of artificial intelligence has allowed for the design of a robust system which can `self-learn' to recognise any new disease samples. Research involved investigation of a number
of camera and hardware options.
Final system validation was conducted on both `stock' disease images provided by agronomists, and on actual plant samples, which proved that the system could function across a broad range of diseases and crops with a degree of accuracy between 95-99%.
This research indicates that it is possible to develop tools which can give an immediate analysis at all stages of infection, and be robust enough to work over a range of diseases and crops. Further development and refinement would provide a useful diagnosis tool for both growers and experts.
|
Statistics for this ePrint Item |
| Item Type: | USQ Project |
|---|---|
| Item Status: | Live Archive |
| Additional Information: | Bachelor of Engineering (Mechatronics) project. |
| Faculty/School / Institute/Centre: | Historic - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 - 31 Dec 2021) |
| Supervisors: | McCarthy, Cheryl |
| Date Deposited: | 09 Sep 2015 05:15 |
| Last Modified: | 09 Mar 2016 04:44 |
| Uncontrolled Keywords: | wheat crop diseases; k-means clustering; gabor filters; artificial neural networks Neural Networks |
| Fields of Research (2008): | 09 Engineering > 0999 Other Engineering > 099901 Agricultural Engineering |
| Fields of Research (2020): | 40 ENGINEERING > 4099 Other engineering > 409901 Agricultural engineering |
| URI: | https://sear.unisq.edu.au/id/eprint/27268 |
Actions (login required)
 |
Archive Repository Staff Only |
