Improving Generalization Performance of YOLOv8 for Camera Trap Object Detection