Project Number: 8042-42000-021-011-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Aug 1, 2021
End Date: Jul 31, 2024
Meatpacking facilities rely on the visual naked-eye inspection of carcasses. This process is prone to errors. Rapid, user-friendly methods and new devices that can provide greater portability and flexibility in usage are needed for contamination inspection in meat processing facilities. To 1) develop fluorescence-based contaminant detection methodology for Contamination and Sanitation Inspection (CSI) in meat processing facilities; 2) enhance both software and hardware abilities of a handheld CSI device for contamination detection on meat products and equipment surfaces used in meat processing; 3) validate the application of the CSI device as an effective and practical tool in the inspection of meat and meat processing environments for contamination detection.
A newly developed portable handheld imaging device for contamination and sanitation inspection will be used to record images on meat carcasses (beef, pork, and bison) from five different meat processing plants in North Dakota and South Dakota. Meat samples contaminated with feces will also be tested on-site in meat processing plants. A machine learning-based framework will be developed to label (annotate) these fluorescence images. We will construct 500 annotations of fluorescence images. These annotated images can be amplified by automated processes (e.g., rotation, scaling, contrast, etc.) to produce 20,000 - 50,000 images for artificial intelligence (AI) machine learning. We will develop image processing algorithms to detect contaminants on meat products and equipment surfaces. We then will evaluate and validate the CSI device performance for the detection of fecal residues or other contamination on carcasses and processing equipment on-site in meat processing plants in North Dakota and South Dakota.