Skip to main content

pipeline

logger_config​

logger​

VAMEPipeline Objects​

class VAMEPipeline()

VAME pipeline class.

__init__​

def __init__(project_name: str,
             poses_estimations: List[str],
             source_software: Literal["DeepLabCut", "SLEAP", "LightningPose"],
             working_directory: str = ".",
             videos: Optional[List[str]] = None,
             video_type: str = ".mp4",
             fps: Optional[float] = None,
             copy_videos: bool = False,
             paths_to_pose_nwb_series_data: Optional[str] = None,
             config_kwargs: Optional[dict] = None) -> None

Initializes the VAME pipeline.

Parameters

  • project_name (str): Project name.
  • videos (List[str]): List of video files.
  • poses_estimations (List[str]): List of pose estimation files.
  • source_software (Literal["DeepLabCut", "SLEAP", "LightningPose"]): Source software used for pose estimation.
  • working_directory (str, optional): Working directory, by default ".".
  • video_type (str, optional): Video file type, by default ".mp4".
  • fps (int, optional): Sampling rate of the videos. If not passed, it will be estimated from the video file. By default None.
  • copy_videos (bool, optional): Copy videos, by default False.
  • paths_to_pose_nwb_series_data (Optional[str], optional): Path to pose NWB series data, by default None.
  • config_kwargs (Optional[dict], optional): Additional configuration keyword arguments, by default None.

Returns

  • None

get_states​

def get_states(summary: bool = True) -> dict

Returns the pipeline states.

Returns

  • dict: Pipeline states.

get_sessions​

def get_sessions() -> List[str]

Returns a list of session names.

Returns

  • List[str]: Session names.

get_raw_datasets​

def get_raw_datasets() -> xr.Dataset

Returns a xarray dataset which combines all the raw data from the project.

Returns

  • dss (xarray.Dataset): Combined raw dataset.

preprocessing​

def preprocessing(centered_reference_keypoint: str = "snout",
                  orientation_reference_keypoint: str = "tailbase") -> None

Preprocesses the data.

Parameters

  • centered_reference_keypoint (str, optional): Key point to center the data, by default "snout".
  • orientation_reference_keypoint (str, optional): Key point to orient the data, by default "tailbase".

Returns

  • None

create_training_set​

def create_training_set(
        test_fraction: float = 0.1,
        split_mode: Literal["mode_1", "mode_2"] = "mode_1") -> None

Creates the training set.

Parameters

  • test_fraction (float): Test fraction.
  • split_mode (str, optional): Split mode, by default "mode_1".

Returns

  • None

train_model​

def train_model() -> None

Trains the model.

Returns

  • None

evaluate_model​

def evaluate_model() -> None

Evaluates the model.

Returns

  • None

run_segmentation​

def run_segmentation() -> None

Runs the pose estimation segmentation into motifs.

Returns

  • None

run_community_clustering​

def run_community_clustering() -> None

Runs the community clustering.

Returns

  • None

generate_motif_videos​

def generate_motif_videos(
        video_type: str = ".mp4",
        segmentation_algorithm: Literal["hmm", "kmeans"] = "hmm") -> None

Generates motif videos.

Parameters

  • video_type (str, optional): Video type, by default ".mp4".
  • segmentation_algorithm (Literal["hmm", "kmeans"], optional): Segmentation algorithm, by default "hmm".

Returns

  • None

generate_community_videos​

def generate_community_videos(
        video_type: str = ".mp4",
        segmentation_algorithm: Literal["hmm", "kmeans"] = "hmm") -> None

Generates community videos.

Parameters

  • video_type (str, optional): Video type, by default ".mp4".
  • segmentation_algorithm (Literal["hmm", "kmeans"], optional): Segmentation algorithm, by default "hmm".

Returns

  • None

generate_videos​

def generate_videos(
        video_type: str = ".mp4",
        segmentation_algorithm: Literal["hmm", "kmeans"] = "hmm") -> None

Generates motif and community videos.

Parameters

  • video_type (str, optional): Video type, by default ".mp4".
  • segmentation_algorithm (Literal["hmm", "kmeans"], optional): Segmentation algorithm, by default "hmm".

Returns

  • None

visualize_preprocessing​

def visualize_preprocessing(scatter: bool = True,
                            timeseries: bool = True,
                            show_figure: bool = False,
                            save_to_file: bool = True) -> None

Visualizes the preprocessing results.

Parameters

  • scatter (bool, optional): Visualize scatter plot, by default True.
  • timeseries (bool, optional): Visualize timeseries plot, by default True.
  • show_figure (bool, optional): Show the figure, by default False.
  • save_to_file (bool, optional): Save the figure to file, by default True.

Returns

  • None

visualize_model_losses​

def visualize_model_losses(save_to_file: bool = True,
                           show_figure: bool = True) -> None

Visualizes the model losses.

Parameters

  • save_to_file (bool, optional): Save the figure to file, by default False.
  • show_figure (bool, optional): Show the figure, by default True.

Returns

  • None

visualize_motif_tree​

def visualize_motif_tree(
        segmentation_algorithm: Literal["hmm", "kmeans"]) -> None

Visualizes the motif tree.

Parameters

  • segmentation_algorithm (Literal["hmm", "kmeans"]): Segmentation algorithm.

Returns

  • None

visualize_umap​

def visualize_umap(label: Literal["community", "motif"] = "community",
                   segmentation_algorithm: Literal["hmm", "kmeans"] = "hmm",
                   show_figure: bool = False,
                   save_to_file: bool = True) -> None

Visualizes the UMAP plot.

Parameters

  • label (Literal["community", "motif"], optional): Label to visualize, by default "community".
  • segmentation_algorithm (Literal["hmm", "kmeans"], optional): Segmentation algorithm, by default "hmm".

Returns

  • None

report​

def report(segmentation_algorithm: Literal["hmm", "kmeans"] = "hmm") -> None

Generates the project report.

Parameters

  • segmentation_algorithm (Literal["hmm", "kmeans"], optional): Segmentation algorithm, by default "hmm".

Returns

  • None

run_pipeline​

def run_pipeline(from_step: int = 0,
                 preprocessing_kwargs: dict = {},
                 trainingset_kwargs: dict = {}) -> None

Runs the pipeline.

Parameters

  • from_step (int, optional): Start from step, by default 0.
  • preprocessing_kwargs (dict, optional): Preprocessing keyword arguments, by default .
  • trainingset_kwargs (dict, optional): Training set keyword arguments, by default .

Returns

  • None

unique_in_order​

def unique_in_order(sequence)