Image or video inference with detection, segmentation, pose, oriented bounding boxes, and classification using the latest Ultralytics YOLO12 models.