A Novel YOLOv6 Object Detector for Monitoring Piling Behavior of Cage-Free Laying Hens

Piling behavior (PB) is a common issue that causes negative impacts on the health, welfare, and productivity of flock in poultry houses (e.g., cage-free layer, breeder, broiler). Birds pile top each other, weight birds can cause physical injuries, such as bruising or suffocation, may even result death. In addition, PB stress anxiety birds, leading to reduced immune function increased susceptibility disease. Therefore, piling has been reported one most concerning production issues layer houses. Several strategies adequate space, environmental enrichments, genetic selection) have proposed prevent mitigate laying hens, but less scientific information available control it so far. The current study aimed develop test performance novel deep-learning model for detecting evaluate its effectiveness four CF hen facilities. To achieve this goal, utilized different versions YOLOv6 models YOLOv6t, YOLOv6n, YOLOv6s, YOLOv6m, YOLOv6l, YOLOv6l relu). objectives were reliable efficient tool commercial egg-laying facilities based deep learning new research used dataset comprising 9000 images 6300 training, 1800 validation, 900 testing). results show relu-PB perform exceptionally well with high average recall (70.6%), [email protected] (98.9%), [email protected]:0.95 (63.7%) compared other models. detection increases when camera placed close areas. Thus, newly developed demonstrated superior given tested