Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: On-Line Automated Inspection of Poultry Carcasses by Machine Vision

Authors
item Chen, Yud
item Chao, Kuanglin
item Hruschka, William

Submitted to: Computers in Agriculture International Conference Proceedings
Publication Type: Proceedings
Publication Acceptance Date: March 12, 2001
Publication Date: N/A

Interpretive Summary: Development of an automated poultry inspection system that is low-cost, operates with minimum human intervention, and is able to maintain its accuracy is important to the FSIS and the poultry industry. Such systems, placed strategically in the processing plants, would help improve the inspection speed, minimize problems of human error and variability, improve the effectiveness of the federal inspection program, and increase the slaughter plants' productivity. This paper briefly reviews the work conducted at the Instrumentation and Sensing Laboratory (ISL) to apply machine vision technology to develop such a system. The results of the testing of the automatic machine vision inspection system at a commercial poultry processing plant are detailed. This information is important to the regulatory agencies, such as FSIS and FDA, to poultry processing companies, poultry equipment manufacturers, and to the engineers and scientists interested in developing any systems for inspecting agricultural products for safety and quality.

Technical Abstract: Development of an automated poultry inspection system that is low-cost, operates with minimum human intervention, kand is able to maintain its accuracy is important to the U.S. Food Safety and Inspection Service (FSIS) and the poultry industry. Such a system, placed strategically in the processing plants, would help improve the inspection speed, minimize problems of human error and variability, improve the effectiveness of the federal inspection program, and increase the slaughter plants' productivity. This paper describes an automated poultry inspection system that utilizes machine vision technology. The system was tested on-line at a poultry plant. The results showed that the classification accuracy for wholesome and unwholesome carcasses was 94% and 87%, respectively.

Last Modified: 7/28/2014