Instalation

pip install automlhelper

Usage

See example.py

Class Documentation

AutoMLHelper(self, bucket_name, csv_path, project_name)

Args: bucket_name (int): The name of the bucket utilized in AutoML e.g.: my-bucket-vcm csv_path (str): The path where yours CSVs with reference of images and classes of your images. e.g.: csv/ project_name (str): Your Google Cloud project

tag_performance

AutoMLHelper.tag_performance(self)

Returns tag display_name and performance object.

Warning: This connection between tag display_name and the performance object is not returned by Google`s API, this method utilizes an alternative way that can stop working at any moment.

Returns: list: (display_name, performance_object)

tag_and_performance

AutoMLHelper.tag_and_performance(self, key='base_au_prc')

Returns tag display_name and performance metric.

Args: key (str): performance metric of performance object defaults: 'base_au_prc'

Returns: list: (display_name, performance_metric)

best_performing

AutoMLHelper.best_performing(self, key='base_au_prc', limit=0, reverse=True)

Returns best or worst tags by performance metric.

Args: key (str): performance metric of the performance object defaults: 'base_au_prc' limit (int): max number of results reverse (bool): True for best performing, False for worst performing

Returns: list: (display_name, performance_metric)

get_tags

AutoMLHelper.get_tags(self)

Returns all tags in the last CSV.

Returns: list: display_name

get_examples_of

AutoMLHelper.get_examples_of(self, tag_name, limit=10)

Returns examples of tag_name in the last CSV.

Args: tag_name (str): tag display_name limit (int): max number of results

Returns: list: files marked with that tag

predict_gs

AutoMLHelper.predict_gs(self, gs_file)

Returns AutoML predictions on gs_file.

Args: gs_file (str): image in the gs bucket e.g.: gs://my-bucket-name/image.jpg

Returns: list: (display_name, classification score)