<![CDATA[Temperature and humidity readings can show changes inside a quail cage, but they do not tell us how the birds are responding to those conditions. This study developed a low-cost monitoring system that combines environmental sensing with image-based analysis of quail behaviour. A DHT22 sensor measured cage temperature and humidity, while an ESP32 collected and transmitted the readings. A Raspberry Pi 4 processed images captured by a USB camera and ran a Convolutional Neural Network (CNN) based on MobileNet to classify quail behaviour into three operational categories: Normal, Stress, and Aggressive. Environmental readings and classification results were sent through MQTT, stored in a MySQL database, and displayed together on a web dashboard. The system also generated an alert when an abnormal condition was detected. In five comparisons with reference instruments, the DHT22 showed average errors of 0.36% for temperature and 0.38% for humidity. The MobileNet model reached an aggregate validation accuracy of 91.3% on 1,200 labelled images. The average time from image capture to alert delivery was 4.7 seconds. These results show that environmental data and behavioural classification can be processed together on a Raspberry Pi-based platform under the tested conditions. The system was developed for quail monitoring, but its combination of sensors, real-time data, IoT communication, and image classification could also be used as a practical context for interdisciplinary learning.]]>
