Improving Temperature Measurement Accuracy for Layer Chickens Using Depth Estimation with Visible Light and Low-Cost Thermal Infrared Cameras

Layer chicken farming is a vital component of modern agriculture, offering sustainable protein and supporting local economies. Accurate temperature monitoring is critical for early disease detection and maintaining optimal conditions, enhancing productivity and sustainability. This study proposes a novel framework for precise, non-contact temperature measurement in layer chickens using visible light and low-cost thermal infrared cameras. The framework improves temperature measurement precision by calibrating heterogeneous cameras for depth estimation and reducing depth-related errors. It integrates a depth-aware feature matching algorithm and a depth-based correction scheme to adjust temperature readings based on measured depth. Experiments demonstrate that the framework maintains depth errors within 0.15 m, reducing temperature measurement errors from 3 °C to 0.5 °C, well below the 2 °C fever detection threshold. Additionally, under stable environmental conditions, the framework decreases temperature discrepancies between individuals by over 60%, significantly improving anomaly detection rates. This cost-effective solution for multi-point temperature monitoring holds promise for applications in agriculture, industry, and public health.