Detecting sow vulva size change around estrus using machine vision technology

Accurate estrus detection of sows is critical to achieving a high farrowing rate and maintaining good repro-ductive performance. The conventional method of estrus detection uses a back pressure test by breeding tech-nicians, which is time-consuming and labor-intensive. This study aimed to develop an automated estrus detection method by monitoring the change in vulva swelling around the estrus using a LiDAR camera. A total of seven multiparous individually housed sows and a gilt were monitored once per day for 19 consecutive days, starting from 2 days before they stopped receiving Matrix (R). A three-dimensional (3D) point cloud of the vulva region was manually acquired using the LiDAR camera at 0.7 -1.0 m from the back of the sows. The accuracy of the LiDAR camera was examined in a laboratory before imaging sows. Results showed that the measurement error in depth was 3.4 +/- 3.0 mm (mean +/- SD). Collected point cloud data of sows were processed using a customized algorithm to create 3D models of the vulva region by separating them from the sow's body. Five 2D and 3D features were extracted from the 3D models to describe vulva size. Linear regression analysis showed that the calculated volume (CV = width x length x height) could represent the vulva volume (R2 = 0.92). Results also showed that the vulva volume was a reliable estrous indicator. Swelling duration and intensity showed large variation among different sows. This study also indicates that sows with larger vulva volume had a smaller percentage of increase around estrus. The results suggest that the LiDAR camera has the potential as a non-invasive tool to help identify the sow's estrus.