Skip to content

deeplabcut.modelzoo.webapp.inference

Classes:

Name Description
SingletonTopDownRunners

Singleton class for topdown runners.
SuperanimalPyTorchInference

Superanimal inference class.

SingletonTopDownRunners

Singleton class for topdown runners.

This class is a singleton class for topdown runners. It is used to ensure that only one instance of the topdown runners is created.

Attrs

config: Configuration dictionary pose_model_path: Path to the pose model detector_model_path: Path to the detector model num_bodyparts: Number of bodyparts max_individuals: Maximum number of individuals

Source code in deeplabcut/modelzoo/webapp/inference.py 
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
class SingletonTopDownRunners:
    """Singleton class for topdown runners.

    This class is a singleton class for topdown runners. It is used to
    ensure that only one instance of the topdown runners is created.

    Attrs:
        config: Configuration dictionary
        pose_model_path: Path to the pose model
        detector_model_path: Path to the detector model
        num_bodyparts: Number of bodyparts
        max_individuals: Maximum number of individuals
    """

    _instance = None

    def __new__(cls, *args, **kwargs):
        if not cls._instance:
            cls._instance = super().__new__(cls)
        return cls._instance

    def __init__(
        self,
        config,
        pose_model_path: str,
        detector_model_path: str,
        num_bodyparts: int,
        max_individuals: int,
    ):

        pose_runner, detector_runner = get_inference_runners(
            config,
            snapshot_path=pose_model_path,
            max_individuals=max_individuals,
            num_bodyparts=num_bodyparts,
            num_unique_bodyparts=0,
            detector_path=detector_model_path,
        )
        self.pose_runner = pose_runner
        self.detector_runner = detector_runner

SuperanimalPyTorchInference

Superanimal inference class.

This class is used to perform inference on a superanimal model from the DeepLabCut model zoo website.

Source code in deeplabcut/modelzoo/webapp/inference.py 
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
class SuperanimalPyTorchInference:
    """Superanimal inference class.

    This class is used to perform inference on a superanimal model from the DeepLabCut
    model zoo website.
    """

    def __init__(
        self,
        project_name: str,
        pose_model_type: str = "hrnet_w32",
        detector_model_type: str = "fasterrcnn_resnet50_fpn_v2",
        max_individuals: int = 30,
        device: str = "cpu",
    ):
        self.max_individuals = max_individuals
        config = modelzoo.load_super_animal_config(
            super_animal=project_name,
            model_name=pose_model_type,
            detector_name=detector_model_type,
        )
        config = update_config(config, max_individuals, device)
        self._config = config

    def initialize_models(self, pose_model_path: str, detector_model_path: str):
        self.models = SingletonTopDownRunners(
            self.config,
            pose_model_path,
            detector_model_path,
            len(self.config["bodyparts"]),
            self.max_individuals,
        )

    @property
    def config(self):
        return self._config

    def predict(self, frames: dict[str, np.array]):

        input_images = np.array(list(frames.values()), dtype=float)

        bbox_predictions = self.models.detector_runner.inference(images=input_images)
        input_images = list(zip(input_images, bbox_predictions, strict=False))
        predictions = self.models.pose_runner.inference(images=input_images)
        predictions = [{("markers" if k == "bodyparts" else k): v for k, v in d.items()} for d in predictions]
        predictions = [{**item[1], "image_path": item[0]} for item in zip(frames.keys(), predictions, strict=False)]
        responses = {
            "joint_names": self.config["bodyparts"],
            "predictions": predictions,
        }

        return responses