Pig cough sound monitoring and early warning system based on deep learning

Abstract

The invention discloses a pig cough sound monitoring and early warning system based on deep learning, which comprises a data acquisition module, a route, a deep learning server, a cloud server and anearly warning client, and audio information acquired in a pig house is acquired and stored through a pickup; local audio data is uploaded to a deep learning server in real time by using a Socket wireless communication technology, and preprocessing and feature acquisition are performed on pig house sound signals in a complex environment by using an acoustic analysis technology; pig cough sound change rules is researched and compared based on a BLSTM (Bidirectional Long Short Term Memory) and a DFSMN (Deep Feedforward Sequence Memory Neural Network), and an end-to-end cough sound recognition model is established; finally, a pig respiratory system disease early warning platform is formed, and pig respiratory system disease early warning is efficiently and accurately achieved.

Description

technical field [0001] The invention relates to the field of intelligent breeding, in particular to a pig cough sound monitoring and early warning system based on deep learning. Background technique [0002] The pig industry is an important part of my country's animal husbandry. The output and consumption of pork are more than 50% of other meat. Despite the impact of African swine fever, pork demand still accounts for the largest proportion of animal meat used. However, problems such as high breeding density and poor ventilation in pig houses make respiratory diseases one of the most common and serious diseases. Therefore, early monitoring of porcine respiratory diseases is particularly important. Coughing is one of the main symptoms of porcine respiratory diseases, and porcine respiratory diseases can be detected in time by monitoring the sound of pig coughing. However, the current traditional monitoring methods mainly rely on human beings, which is time-consuming and la...

AI Technical Summary

Problems solved by technology

This solution lacks the function of remote monitoring of pig behavior

[0007] In addition, the existing early warning system for pig cough sound recognition, such as Chinese patent CN110189756A, mostly uses machine learning methods, and the accuracy is low

To sum up: the current pig cough recognition algorithm generally uses machine learning, and the recognition accuracy is low; the pig cough early warning device or system has problems such as low transmission audio rate or low algorithm operation speed

Images