predict_dataset

lightning_pose.utils.predictions.predict_dataset(model: Model, data_module: BaseDataModule, preds_file: str | list[str], cfg: DictConfig | ListConfig | None = None) → pd.DataFrame | dict[str, pd.DataFrame]

Save predicted keypoints for a labeled dataset.

Parameters:

• model – API model wrapper; its underlying lightning module is used for inference.

• data_module – data module that contains dataloaders for train, val, test splits.

• preds_file – path for the predictions .csv file.

• cfg – hydra config; if None, falls back to model.config.cfg.

Returns:

pandas dataframe with predictions or dict with dataframe of predictions for each view