Correct oestrous detection is essential to optimise sows’ reproductive efficiency. The traditional technique of oestrous detection depends on the laborious back-pressure check.
Goal: This examine presents an automatic oestrous detection technique for sows housed in particular person stalls utilizing a robotic imaging system and neural networks.
Strategies: A robotic imaging system consisting of a LiDAR digicam was used to observe a gaggle of stall-housed sows at a 10-min interval to seize their postures and vulva quantity. Imagery information have been analysed utilizing a beforehand developed pipeline.
Outcomes: Outcomes confirmed that important modifications have been noticed in each day standing index, sternal mendacity index, lateral mendacity index, posture change frequency, and vulva quantity earlier than the onset of oestrous. A 1-D convolutional neural community mannequin structure for oestrous detection was developed utilizing Days from Weaning, behaviour options, and vulva quantity options as inputs. The oestrous detection fashions have been evaluated utilizing 10-fold cross validation. The coaching and testing accuracies of the oestrous detection mannequin have been 96.1 ± 2.0% and 92.3 ± 10.1% when utilizing the Days from Weaning and behavior options as enter. The mannequin’s coaching and testing accuracies elevated to 98.1 ± 2.4% and 98.0 ± 4.2% when vulva quantity options have been added to the enter.
Conclusion: Whereas it’s troublesome to hint the behaviour of sows housed in group circumstances, combining vulva quantity options with Days from Weaning might be an appropriate technique to detect the onset of oestrous in these sows. The coaching and testing accuracies of this technique of oestrous detection have been 97.9 ± 1.4% and 95.2 ± 4.8%. Nonetheless, additional validation below actual group home circumstances is required.
Ziteng Xu, Jianfeng Zhou, Corinne Bromfield, Teng Teeh Lim, Timothy J. Safranski, Zheng Yan, Jeffrey G. Wiegert. Automated oestrous detection in sows utilizing a robotic imaging system. Biosystems Engineering. 2024; 244: 134-145. https://doi.org/10.1016/j.biosystemseng.2024.05.018.
