CNN-Based Sheep Vocalization Analysis for Emotional State Recognition and Health Monitoring

The emotional state of sheep is closely related to their health status, traditional monitoring methods rely on manual observation with low efficiency, this study proposes a convolutional neural network-based sheep vocalization analysis method for emotional state recognition and health monitoring, by collecting sheep vocalization samples under different emotional states and extracting Mel-frequency cepstral coefficient features, a deep convolutional neural network model is constructed for classification recognition, experimental results show that this method achieves 94.3% accuracy in four emotional state classification tasks, with a recall rate of 93.8%, F1-score of 94.0%, improving by 12.7% compared to traditional machine learning methods, this system can monitor sheep flock health status in real-time and detect abnormal individuals promptly, providing technical support for intelligent livestock farming.