AI Pig Health Monitoring Improved to Reflect Farm Feedback… Early Detection of Piglet Crushing and Coughs
Chungnam National University Professor Jung Sun-ok’s team holds second on-site briefing in Buyeo… housing and Korean-language menus improved With the research ending this month, the challenge is linking it to follow-up projects
AI Briefing
On the 2nd, a research team led by Professor Jung Sun-ok of Chungnam National University unveiled an intelligent pig health monitoring system in Buyeo with improved durability and ease of use. Using RGB and thermal imaging cameras and sound sensors, the system detects early warning sounds of piglet crushing risk, coughing, and fever, and is being demonstrated at pig farms in the Chungcheong and Jeolla regions. As the research ends this month, securing demonstration farms and follow-up projects is considered a key task.
A pig barn is a space where it is difficult for people to remain all day. Piglet crushing deaths caused by being pinned under sows, or respiratory diseases that begin with coughing, occur in brief moments or show only small early signs, making them easy to miss through visual management alone.
According to the ‘2025 Computerized Performance Results of Handon Farms Nationwide Korean Pork Farms’ released in May by the Handon Research Institute of the Korean Pork Checkoff Management Committee, among professional Handon Farms user farms, the top 10% in productivity recorded an MSY(annual number of pigs shipped per sow) of 26.2 head, while the bottom 10% recorded 13.7 head, a difference of 12.5 head. Converted to a farm with 100 sows, the annual number of pigs shipped differs by 1250 head.
Applying the average October farm-gate price for finishing pigs(per head based on 110㎏ 530, 3000 won) compiled by the Korea Institute for Animal Products Quality Evaluation, the annual sales gap reaches about 666 million won. The pre-weaning survival rate, the percentage of piglets that survive before being weaned, also differed sharply: the top 10% recorded 90.6%, while the bottom 10% recorded 76.6%. This is why attention is focusing on technologies that can detect losses occurring in brief moments, such as crushing and disease, at an early stage.
A research team from Chungnam National University, which held its first on-site briefing in July saying it would solve this problem with AI, has presented an improved system reflecting farm feedback at the site again after about three months.
The research team led by Professor Jung Sun-ok of Chungnam National University’s Department of Smart Agricultural Systems and Machinery Engineering held an ‘On-site Demonstration Briefing and Training for an Intelligent Pig Health Monitoring System’ at the Buyeo-gun Agricultural Technology Center on the 2nd. About 20 people attended the event, including Chungnam National University researchers, experts from academia·and government, (incorporated association)farms belonging to the Buyeo branch of the Korea Pork Producers Association.
This project is part of the 2026 ‘Agricultural Technology Industry-Academia Cooperation Support Project’ operated by the Rural Development Administration and implemented by the Korea Agriculture Technology Promotion Agency. The goal is to create prototypes using technology held by the research team, install them on actual pig farms, verify whether they can be used in the field and their potential for utilization, and then lead to practical application and distribution to farms.
The system remotely checks the condition of pigs using RGB cameras, thermal imaging cameras, and sound sensors installed in pig pens. RGB cameras automatically count the number of pigs in each pen and track major behavior patterns such as lying down·standing·walking·and feeding. Thermal imaging cameras measure the maximum·minimum·and average values of pig body surface temperature in real time to identify animals suspected of having fever and risks of heat stress due to high temperatures. Sound sensors detect cough sounds and sounds indicating piglet crushing risk in the 50Hz~16kHz range, as well as abnormal vocalizations such as screams caused by tail biting, and respond within 1.5 seconds.
According to the research team, based on the first demonstration product, the RGB camera’s accuracy in recognizing the number of animals·and behaviors was 95~97%. The sound sensor’s detection accuracy was more than 95% for cough sounds, and more than 90% for crushing-risk sounds.
The collected video and audio data are transmitted to a server via Wi-Fi(Wi-Fi). Farm owners can remotely check the inside of pig pens 24 hours a day through a PC web dashboard or smartphone app, and if an emergency situation such as an abnormal sound or fever occurs, they immediately receive a warning notification on their mobile phone.
The focus of this briefing was ‘what was fixed in the field’. The research team further strengthened the housing(enclosure)that contains the equipment so it can withstand dusty and humid livestock barn environments, and improved durability by reworking connection parts such as connectors so they do not loosen. For ease of use, compared with the first product, the menus were changed to Korean so farmers could understand them more easily, and the menu structure was revised so that multiple pieces of information can be viewed together on one screen. On the day, the research team demonstrated the process of checking data coming from pig-pen devices in real time on the user screen, and listened to opinions from participating farms on what is needed for industrialization and farm distribution.
The demonstration is being conducted separately in two regions. Chungnam National University, the lead research and development institution, is in charge of the Chungcheong region, while the team led by Professor Lee Su-hyeop of Korea National University of Agriculture and Fisheries, a joint research institution, is in charge of the Jeolla region, demonstrating the same system at pig farms in each region. Farmers’ user experiences and opinions were reflected in improving the system’s performance and field usability.
The remaining task is practical application. The research period runs through this month, and this briefing was held in the final stage of the project. The research team plans to pursue commercialization and distribution to farms in the future, but there are considerable hurdles to overcome before research carried out as a government project can lead to follow-up projects and be distributed in the field. Even finding livestock barns for demonstrations is difficult. This is because pig farms concerned about disease introduction do not readily open their farms to outsiders. Once the practical application stage begins, funding issues will also follow.
Professor Jung said, “This project became an opportunity to identify farm demand in concrete terms”. She added, “I hope there will be government interest along with support for R&Dor practical application”, emphasizing, “I hope government support will continue through practical application and all the way to expansion among farms”.
Reporter Lee Hyun-woo leehw@agtechkorea.com
This article has been automatically translated by AI (Artificial Intelligence).
Comments
0