|MENDOZA, QUERRIEL - Kansas State University|
|NIELSEN, MITCHELL - Kansas State University|
|CAMPBELL, JAMES - US Department Of Agriculture (USDA)|
Submitted to: Artificial Intelligence
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/3/2023
Publication Date: 3/15/2023
Citation: Mendoza, Q.A., Pordesimo, L.O., Nielsen, M.L., Armstrong, P.R., Campbell, J.F. 2023. Application of machine learning for insect monitoring in grain facilities. Artificial Intelligence. 4:348-360. https://doi.org/10.3390/ai4010017.
Interpretive Summary: Stored product insects are small and are difficult to locate and identify to species level when they infest large food facilities and storage structures. While several techniques exist for monitoring for stored product insect populations, the methods can be labor intensive and certain species may need to be sent to taxonomists for identification. For this study, a low-cost camera-based insect detection system was tested that could take images of moving insects in real time. In addition, artificial intelligence (AI)-based models were developed and trained to detect the presence of insects and identify the species. Because light conditions can vary in the field, we also tested our camera system and AI model using either a white LED light, yellow LED light, or no lights. Overall, the model performed well and was able to correctly detect and identify warehouse and cigarette beetles 81% of the time. Therefore, this methodology shows promise for being a rapid, affordable, and effective solution for monitoring for and identifying stored grain pests in the field. Moving forward, we are exploring the use of an autofocus camera and expanding the breadth of our dataset to increase the accuracy and robustness of the model.
Technical Abstract: A basic insect detection system consisting of a manual focus camera, an AI computer, and a trained deep learning model was developed. The model was validated through a live visual feed. Being able to detect, classify and monitor insect pests present in a grain storage or food facility in real-time will be vital to decision making about insect control. The camera captures the image of the insect and passes it to a Jetson Nano for processing. The Jetson Nano runs the trained deep learning model to detect the presence and species of insects. Using three different lighting situations: white LED light, yellow LED light, and no lighting condition, the results of the detection are displayed on a monitor. Validating using F1 scores and comparing the accuracy based on lights sources, the system was tested with a variety of stored grain insect pests and was able to accurately detect the presence and type of insect. The results demonstrate that the system is an effective and affordable automated solution to insect detection. Such an automated insect detection system can help reduce pest control costs and save producers time and energy while safeguarding the quality of stored products.