A groundbreaking technology has been developed that enables artificial intelligence (AI) to determine if pigs are pregnant—a task traditionally reserved for experts. This innovation allows even non-experts to assess pig pregnancies, potentially boosting farm production efficiency easily.
Today, South Korea’s Rural Development Administration (RDA) announced that it had developed the world’s first AI technology for detecting pig pregnancies and filed three patent applications.
Traditionally, confirming pig pregnancies involved observing sow behavior starting 21 days after artificial insemination and interpreting ultrasound images of the uterus after 25 days when gestational sacs become more visible. The accuracy and timing of these detections relied heavily on the interpreter’s skill in reading ultrasound images.
To improve this process, the RDA trained the AI system using over 430,000 ultrasound images of pig uteri. The AI analyzes abdominal ultrasound images captured over more than 10 seconds to determine and report the pregnancy status.
This new technology has achieved over 95% accuracy in detecting pregnancies between 22 and 25 days after artificial insemination. By accelerating pregnancy detection, the system increases the health management period for pregnant pigs, which is expected to enhance productivity. Also, it helps reduce non-productive days by facilitating timely re-insemination for un-pregnant sows, leading to cost savings on feed.
The RDA aims to refine the AI model to achieve over 95% accuracy in detecting pregnancies between 18 and 21 days before the typical estrus cycle.
Im Gi Soon, director of the National Institute of Animal Science at the RDA, commented, “In addition to pregnancy detection, we are developing various AI technologies for managing sow body condition and detecting risks for piglets. Our goal is to expand smart farm technologies to increase productivity, reduce farming costs, and enhance the competitiveness of pig farms.”
%20to%20determine%20if%20pigs%20are%20pregnant.){kind=link}