Oestrus Analysis of Sows Based on Bionic Boars and Machine Vision Technology.

摘要: This study proposes a method and device for the intelligent mobile monitoring of oestrus on sow farm, applied in field production. A bionic boar model that imitates sounds, smells, touch real boars was built to detect sows after weaning. Machine vision technology used identify interactive behaviour between empty establish deep belief network (DBN), sparse autoencoder (SAE), support vector machine (SVM) models, resulting recognition accuracy rates were 96.12%, 98.25%, 90.00%, respectively. The interaction times frequencies static behaviours both ears during heat further analysed. results show there is strong correlation duration contact ears. average 29.7 s/3 min, which remained 41.3 min. interactions as basis judging sow’s states. In contrast with methods other studies, proposed innovative design recyclable can be check emotions, quickly behaviours. approach more accurately obtain provide scientific reference conception time.