Design of a Prototype Monitoring System for Poultry Security using IoT and Machine Learning

In recent years, animal welfare has become very popular to reduce the effects of livestock living in small spaces. Free-range chicken farms are one of the ways to treat chickens with care and less stress. Several recent technologies were employed to improve animal wellness and product quality. Internet of Things and Machine Learning were widely implemented in many applications. This paper presents the prototype design of the monitoring system for poultry security. Two types of chicken sounds, alarm and infection sounds, were used to create a classification model based on Google’s Teachable Machine. Three types of ESP32-based RFID reader nodes track chickens in the coop. All information and classification data were transmitted to the monitoring dashboard using the MQTT protocol. The experiment showed that the proposed concept is feasible for farms with 93.33% classification accuracy and 100% status monitoring.