Function keypoint_matching

deeplabcut/utils/pseudo_label.py:144–339

Runs the keypoint matching algorithm for a DeepLabCut project. Matches project keypoints to SuperAnimal keypoints automatically, by running SuperAnimal inference on all images in the dataset Args: config_path: The path of the DeepLabCut project configuration file. super

(
    config_path: str | Path,
    superanimal_name: str,
    model_name: str,
    detector_name: str,
    copy_images: bool = False,
    device: str | None = None,
    train_file: str = "train.json",
)

def keypoint_matching(
    config_path: str | Path,
    superanimal_name: str,
    model_name: str,
    detector_name: str,
    copy_images: bool = False,
    device: str | None = None,
    train_file: str = "train.json",
):
    """Runs the keypoint matching algorithm for a DeepLabCut project.

    Matches project keypoints to SuperAnimal keypoints automatically, by running
    SuperAnimal inference on all images in the dataset

    Args:
        config_path: The path of the DeepLabCut project configuration file.
        superanimal_name: SuperAnimal dataset with which to run keypoint matching.
        model_name: Name of the SuperAnimal pose model architecture with which to run
            keypoint matching
        detector_name: Name of the SuperAnimal detector architecture with which to run
            keypoint matching
        copy_images: When False, symlinks are created for the dataset used for keypoint
            matching. Otherwise, images are copied from the `labeled-data` folder to the
            folder used for keypoint matching.
        device: The device on which to run keypoint matching.
        train_file: The name of the file containing the labels to output.
    """
    config_path = Path(config_path)
    cfg = af.read_config(str(config_path))
    dlc_proj_root = config_path.parent

    if "individuals" in cfg:
        temp_dataset = MaDLCDataFrame(str(dlc_proj_root), "temp_dataset")
        max_individuals = len(cfg["individuals"])
    else:
        temp_dataset = SingleDLCDataFrame(str(dlc_proj_root), "temp_dataset")
        max_individuals = 1

    memory_replay_folder = dlc_proj_root / "memory_replay"
    temp_dataset.materialize(str(memory_replay_folder), framework="coco", deepcopy=copy_images)

    # run inference on the train set
    config = modelzoo.load_super_animal_config(
        super_animal=superanimal_name,
        model_name=model_name,
        detector_name=detector_name,
    )
    if device is None:
        device = select_device()

    # get the SuperAnimal detector and pose model snapshot paths
    pose_model_path = modelzoo.get_super_animal_snapshot_path(
        dataset=superanimal_name,
        model_name=model_name,
    )
    detector_path = modelzoo.get_super_animal_snapshot_path(
        dataset=superanimal_name,
        model_name=detector_name,