Background

The application of machine learning within the GLAM sector is of increasing interest. We think it’s important that the application of machine learning isn’t only focused on novel research applications or central tasks like layout-recognition / Optical Character Recognition, but also used for mundane applications. flyswot is focused on exploring one of these more ‘mundane’ machine learning applications.

As part of this more ‘mundane’ application of machine learning, a command line tool has a numbr of nice features:

• It brings the model to the user. They can run the model on the device they use for their work. This hopefully helps demistify the process

• Command Line Applications are already used for other tasks by HMD so there is more potential to fit flyswot into existing workflows

• flyswot is focused on trying to also highlight important information about the model and the models performance to end-users.