Abstract: Chicken activity level is an important indicator reflecting their health．By monitoring activity level of chicken flock, potential health issues can be identified in time to prevent disease spread．Therefore, an anomaly monitoring method for chicken flock activity based on YOLOv8s and BoT-SORT algorithm was proposed．Cameras were used to capture chicken activity videos, YOLOv8s was used to extract appearance and motion features of chickens, and BoT-SORT algorithm was integrated to achieve multi-object tracking．By analyzing movement trajectories, activity level of each chicken was quantified, and an automatic comparison with preset activity thresholds was done to provide timely alerts for abnormal conditions．Experimental results showed that method achieved a multi-object tracking precision（MOTP）of 94.33%, effectively monitoring chicken activity levels and providing early warnings for potential problems．