Behavior detection Algorithm of Caged White-feather broiler based on multi-scale detail feature fusion and object relation inference

To address the issue of chicken behavior detection in large-scale caged environments, this paper proposes an improved object detection algorithm. The algorithm consists of two modules: the multi-scale detailed feature fusion module and the object relationship inference module, which are responsible for extracting detailed features and determining the relationships between objects, respectively. The experimental results show that the proposed improved algorithm achieves the best detection accuracy compared to other state-of-the-art models, both in the COCO dataset and the self-built behavior detection datasets in real large-scale caged white-feather broiler breeding environments. We achieved recognition accuracies of 99.6%, 98.7%, 99.2%, and 98.3% for the most important feeding, drinking, moving, and mouth-opening behaviors that affect the health status of broilers.