image = cv2.imread(img_path)
landmarks, bboxes = torchlm.runtime.forward(image) >> batches

Is there a way to infer in batches using an image list or data loader instead of inferring with a single image?