add_predictions_dataset

CrowdDetectionEvaluator.add_predictions_dataset(predictions_name: str, predictions: Dataset)

Method to add predictions to the Evaluator from a Dataset object. The prediction dataset must match the Evaluator data:

• prediction label_map must be equal or a subset to the

  Evaluator’s label map

• image data must be the same, except the relative path

  (it can change although the image has not) i.e. there must be the same number and ids of images and all the columns in the prediction image data must match the corresponding ones in the evaluator image data.

Note that this method will overwrite a potentially already existing prediction dataframe

Parameters:

• predictions_name – name of predictions to add. It will then be used as key in the self.predictions_dictionary attribute.

• predictions – prediction Dataset, from which the annotations will be extracted and added to the evaluator.