brails.processors.roof_shape_classifier.roof_shape_classifier module

Class object to use and retrain the roof type classifier model.

class brails.processors.roof_shape_classifier.roof_shape_classifier.RoofShapeClassifier(input_data: dict | None = None)

RoofShapeClassifier predicts roof shapes from aerial imagery.

This classifier serves as a wrapper around the ImageClassifier class, setting up the necessary inputs for roof type classification.

model_path (str): Path to the model file. classes (List[str]): List of roof type classes that will be predicted.

predict(images: ImageSet) -> Dict[str, str]:: Returns predictions for the set of images provided.
retrain(data_dir, batch_size=8, nepochs=100, plot_loss=True): Retrains the roof type classifier model with the provided data.

predict(images: ImageSet) → dict

Predict the roof shape for the given images.

dict: A dictionary mapping image keys to their predicted roof: shapes. The keys correspond to the same keys used in ImageSet.images, and the values are the predicted roof shape (Flat, Gable, or Hip).

retrain(data_dir, batch_size=8, nepochs=100, plot_loss=True)

Retrains the roof type classifier model with the provided data.

Args__: data_dir (str): The directory containing training data. batch_size (int): The number of samples per batch. Default is 8. nepochs (int): The number of epochs for training. Default is 100. plot_loss (bool): Whether to plot the loss during training. Default

is True.