State body develops AI-based machine-vision tech to calculate chicken meat weight

SEOUL -- A state agricultural organization joined hands with LG Innotek, the electric component wing of South Korea's LG Group, to develop an artificial intelligence-based machine vision technology that analyzes video images to accurately calculate the weight of chicken meat.

South Korea with some 87,000 fried chicken stores is the home of enthusiastic chicken lovers who consume 936 million chickens annually, about 20 chickens per person. Chickens aged about 30 days, which have an average weight of about one kilogram, (2.2 pounds) are most popular.

Under a contract with franchise operators, farms provide chicken meat based on weight guidelines given by contractors. If the weight of chicken meat exceeds or falls short of the agreement by more than 50 grams, farms would receive penalties of about 200 won ($0.17) per chicken.

The Rural Development Administration (RDA) and LG Innotek have jointly developed a technology that autonomously analyzes and calculates the weight of processed chickens. "In order to develop and industrialize technologies for smart chicken meat farms, we will continue to push for detailed evaluation," said Yang Chang-beom, the head of the National Institute of Animal Science.

RDA said it has created an image database of about 56,000 different chicken meat and actual data such as weight and breeding. Cameras installed in chicken meat farms will examine meat along the processing line with machine-vision technology used at smart farms for quality control. A machine vision system will detect poorly-manufactured products and remove them from the line.