phenocam_snow.predict

classify_online(model, categories, img_url)

Performs online classification.

Parameters:

  • model (PhenoCamResNet) – The model to use.

  • categories (List[str]) – The categories to use.

  • img_url (str) – The URL of the image to run classification on.

Returns:

A 2-tuple where the first element is the image at img_url as a NumPy array and the second element is the predicted label.

classify_offline(model, categories, img_path)

Performs offline classification.

Parameters:

  • model (PhenoCamResNet) – The model to use.

  • categories (List[str]) – The image categories.

  • img_path (str) – The file path of the image to classify.

Returns:

A 2-tuple where the first element is the image at img_path as a NumPy array and the second element is the predicted label.

load_model_from_file(model_path, resnet, n_classes)

Loads a model from checkpoint file.

Parameters:

  • model_path (str) – The path to the model checkpoint file.

  • resnet (str) – The type of Resnet that was used.

  • n_classes (int) – The number of classes.

Returns:

The loaded model.

Return type:

PhenoCamResNet

run_model_offline(model, site_name, categories, img_dir)

Gets predicted labels for all images in a directory.

Parameters:

  • model (PhenoCamResNet) – The model to use.

  • site_name (str) – The name of the PhenoCam site.

  • img_dir (str) – The directory containing the images to classify.

Returns:

A pandas DataFrame with predictions.

Return type:

pd.DataFrame

run_model_online(model, site_name, categories, urls)

Gets predicted label for image online.

Parameters:

  • model (PhenoCamResNet) – The model to use.

  • site_name (str) – The name of the PhenoCam site.

  • url (str) – The URL of the image for which you want a prediction, or the name of a file containing all the URLs, one per line.