Automated Identification of Insect Pests in Traps

Rousseau, Pierre (2020) Automated Identification of Insect Pests in Traps. [USQ Project]

[img]
Preview
Text (Project)
Rousseau_P_Lobsey_Redacted.pdf

Download (10MB) | Preview

Abstract

Insect pests are a significant risk to the agricultural and horticultural sectors and present a major risk to many billions of dollars’ worth of food production. Significant infestations can threaten food security, as well as export markets due to importing countries placing restrictions on a region of origin.

This study investigated the Mediterranean fruit fly (Medfly), species name Ceratitis capitata (Wiedemann). Fruit flies threaten a $12 billion industry and are widespread. Medfly is the most ubiquitous of the fruit fly pest species and therefore poses a significant threat to agriculture worldwide.

The objectives of the project were to develop a low-cost and open source automated fruit fly trap, which could be used for remote monitoring of fruit fly traps. Due to various limitations the system could not be implemented as initially envisaged but a partially functional system was achieved.

Some system components were successfully or partially implemented. Construction and remote operation of a Bosch BME280 temperature, pressure and relative humidity sensor, along with provision for data logging for this sensor was implemented. This system was successfully operated remotely via a WiFi connection. A 16MP camera was partially implemented but failed to perform as expected and further work on this hardware was abandoned. An Adafruit neopixel 12 LED ring was successfully implemented, to be used as a light source for image collection in the dark. Although the camera did not perform as expected, partial success was achieved in processing images by discriminating Medflies within a trap using the OpenCV Simple Blob Detector and saving the fly count to a csv file. No other easily implementable image processing options were found in OpenCV, which could discriminate fruit flies from background. A control algorithm was also developed to drive all the system components and to email a csv file with collated weather data and fly counts, with time stamps, to a designated email address. A housing for the entire system was designed and 3D printed at USQ and a partially operational prototype was constructed.

The viability of the project is still considered to be feasible, although the work required and the challenges involved to achieve completion necessitate that much more time will need to be spent on the project to achieve a fully functional model, with potential for many years of further development.

Obtaining a better camera and further development of the fruit fly identification software, power management, communication protocols and lighting system are all considered to be important.


Statistics for USQ ePrint 43034
Statistics for this ePrint Item
Item Type: USQ Project
Item Status: Live Archive
Additional Information: Bachelor of Engineering (Honours)(Mechatronic)
Faculty/School / Institute/Centre: Current - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 -)
Faculty/School / Institute/Centre: Current - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 -)
Supervisors: Lobsey, Craig
Date Deposited: 12 Aug 2021 01:44
Last Modified: 12 Aug 2021 01:44
URI: http://eprints.usq.edu.au/id/eprint/43034

Actions (login required)

View Item Archive Repository Staff Only