This package contains tools for preparing data for Cloud AI Vision APIs. Currently available:

• Tool to convert a directory of PASCAL VOC files into AutoML dataset CSV.
• Tool to generate an AutoML dataset CSV from nested directory of images.

To install with pip, change to this packages's top directory and run:

pip install .

This package uses Python Fire for handling command-line execution.

Run cvutil the following to get a list of supported commands.

Example: To generate an AutoML Object Detection (OD) CSV from PASCAL VOC files:

cvutil annotations pascal-to-csv <input_dir>  <output CSV file path>

You can find more information about a command by running cvutil <subcommand path> -- --help. Example:

cvutil annotations pascal-to-csv  -- --help

To run all Python unit tests using nose. In the top directory, run:

nosetests

Using unittest:

python -m unittest discover -s cloud_vision_utils -p '*_test.py'

Licensed under the Apache 2.0 License.