Using aerial images and ImageJ to machine count dogfennel (Eupatorium capillifolium) in Florida pastures
Applied Research
J Strickland
County Extension Director
UF/IFAS
Kissimmee
Abstract
Objective: Dogfennel (Eupatorium capillifolium) is a perennial weed commonly occurring in Florida pastures. Dogfennel’s aggressive characteristics have shown to cause significant loss in pasture yield and forage production. Herbicides are the most cost-effective method of control of this weed. Herbicide control assessments such as percentage control estimates and plant counts are methods of evaluating herbicide efficiency. However, percentage control estimates are subjective and require a trained eye, and plants counts are time consuming and labor intensive. The objective of this research was to determine if aerial images and machine counting will provide a more effective and precise plant count compare to hand counting. Methods: Six plots measuring 37.2 m2, were examined using hand counting and machine counting using drone images and ImageJ. ImageJ is a software developed for the national institute of health to count cells in images taken under a microscope. The aerial images were RGB (red, green, blue) and captured using a DJI Phantom 4 at an altitude of 30.5 m. The color threshold used in ImageJ to distinguish the dogfennel from Bahiagrass (Paspalum notatum) was a hue of 55-120, saturation of 129-255, and brightness of 0-167 and a color space of HSB. This research was conducted in July 2022. Results: All plots had an average of 8.83 dogfennel plants when counted by hand. Machine counting indicated that there were on average 8.33 plants. An unpaired t-test was used to analyze the data. The p-value is 0.0577. While not a quite a significant difference, increasing the plot size and repeating this experiment will likely produce a significant difference. In each plot, the hand count number was higher than the machine count number, thus indicating that the aerial image was not adequate for machine counting. Conclusion: One possible reason that the results were not significant was because of the plot size. The other is the inability of the drone to see small plants. A possible remedy for this is to fly the drone at a lower altitude. Machine counting using ImageJ software has shown promise of being a useful tool that we may utilize in the future.
Poster has NOT been presented at any previous NACAA AM/PIC
This poster is being submitted for judging. It will be displayed at the AM/PIC if not selected as a State winner. The abstract will be published in the proceedings.
Click to view Poster
Authors: J Strickland, Jessica Sullivan, Ky Sales, JJ White
-
Strickland, J. County Extension Director, UF/IFAS Extension, Florida, 34744
-
Sullivan, J. Extension Agent III, UF/IFAS Extension, Florida, 34744
-
Sales, K. Extension Agent I, UF/IFAS Extension, Florida, 34461
-
White, J. Soil and Water Service, Osceola County, Florida, 34744