Hi Michael,
What Dennis has presented is really just standard Scalismo, but using some features which are currently poorly documented.
I have created a gist, which should show you how you can customize ScalismoUI:
The following page on the Scalismo webpage might also be useful in this context
The drawing
of the lines in Dennis' demo is implemented using a plugin for ScalismoUI.
For fitting, a standard Metropolis-Hastings algorithm is used as explained in the tutorial. The fitting to the lines is achieved by defining a likelihood function (evaluator) whose log probability is proportional to the distance of the model to the closest point on the line.
Best regards,
Marcel