In the DataProcessing section, we experimented and created some awesome code to download, display, process and extract features from the Saltritus image.
To ensure that improvement made to our code or even updates in the computing environment are not changing the expected results of each one of your modules, we will see how to write basic unit tests.
More details:
Integration testing consists in writting tests that will check all your modules can work together.
To confirm that your commits are not causing failure on other platform or when executed using different version of python, the test suite can automatically be run on a build bot each time an update is pushed to your repository.
The Travis CI platform provides a very convenient solution to trigger a build/test cycle each time a change is pushed on your GitHub repository.
For example: https://travis-ci.org/girder/girder
Then, the associated results could be published on a dashboard.
For example:
Run through the 04-RegressionTesting.ipynb notebook in the repository. You will be able to adapt the example to work with your images. Save the updated notebook to disk.