By this stage you are able to classify your sample of interest in our case the Pycnopodia helianthiodes with the accuracy metric or error metric identified earlier.

Model Deployment

In some cases, the user might want to develop an online application such as a website or application where users in the field such as citizen scientists can be able to use their model for classification. In that case, a comprehensive workflow to achieve this without expert website building skills is outlined in the attached supplementary material.
And just like that, in not so many steps, you have gone through a data science pipeline, AI development and deployment pipeline that would otherwise have cost thousands if not hundreds of thousands of dollars from domain experts.