Fly Monitoring
Farms monitor fly populations using light traps. However, labor shortages prohibit many farms from having fly monitoring programs. Accurate fly monitoring facilitates proper IPM protocols including implementing action thresholds to time pesticide applications. There are two goals of this research: First, we aim to optimize fly attraction to monitoring traps for the most accurate detection of fly populations on farms. Second, we aim to minimize the labor requirement for fly monitoring programs by developing a smartphone application to count and record fly counts from traps.
Optimizing Trap Design
Laboratory studies are ongoing to determine the most attractive wavelength of light to use in light traps. Once optimal wavelengths have been determined, we will test the efficiency of different trap designs. We will also test designs to develop mass trapping stations to control mushroom fly populations on farms.
Computer-vision based fly counting
Labor shortages prevent many farms from implementing fly monitoring programs. Manually counting flies on traps is a labor intensive process. The goal of this research project is to train a convolutional neural network to identify and count mushroom flies on traps. This neural network will be mounted in a smartphone application, which will count and identify flies from pictures that farm workers take of the traps. This work is primarily being done in Dr. Michael Crossley's and Dr. Chandra Kambhamettu's laboratories at the University of Delaware. The PSU Mushroom Fly Research team will assist in field testing of the new technology.
Click here to see more from:
