BEHAVIORAL DYNAMICS AND MOVEMENT PATTERNS IN BROILERS CAPTURED BY MULTI-CAMERA TRACKING SYSTEM

Understanding broiler chicken behavior is critical for advancing precision livestock farming and improving animal welfare, especially under environmental challenges such as heat stress (HS). This study aimed to monitor and quantify three key behaviors in broiler chickens (active, feeding, and drinking) using a multi-view automated tracking system developed by our group. A total of 1,120 male Ross 308 broilers were raised across four production rounds under controlled thermoneutral (TN) and cyclic HS conditions. The experiment employed a custom-built multi-camera system coupled with object detection and tracking algorithms to generate ground-plane tracklets (short segments of a moving object’s trajectory) for individual birds. These tracklets were post-processed to classify behaviors based on movement thresholds and spatial proximity to feeders and drinkers, allowing for a continuous and high-resolution assessment of behavioral dynamics over time. The analysis revealed a progressive decline in general activity levels over time, and of high-intensity movements in particular. In contrast, drinking behavior increased steadily in HS birds. Feeding behavior declined with age and was significantly reduced under HS conditions. Time-of-day effects were also evident: activity peaked in the morning and midday. Additionally, the study tested different radius thresholds to define feeding and drinking zones, finding minimal differences in behavioral outcomes across all radius levels, which supports the robustness of the ethogram. The tracking system achieved high accuracy (Multiple Object Tracking Accuracy: 0.81; ID F1 Score: 0.89), validating its suitability for behavior analysis in group-housed broilers. Few studies have tracked broiler behavior continuously throughout the production cycle, particularly under varying environmental conditions. Our findings help bridge this gap by revealing more information regarding age-related patterns in activity, feeding, and drinking behaviors, including periods of HS. Unlike short-term, event-based studies, our scalable, automated multi-view tracking system enabled long-term behavioral analysis and highlighted the influence of age, HS, and time of day on broiler behavior. With this knowledge of expected behavioral patterns, deviations caused by challenges such as heat stress or disease could be continuously monitored in broilers, enabling 24/7 surveillance and early detection to trigger alerts and support timely interventions. The system’s limitations include the inability to detect complex social behaviors and the constraints of an experimental setting compared to commercial environments