knitr/markdown How to output figures and captions driven by a data set

Question

I\'m trying to produce web pages for a book using figures whose descriptors are in a data frame, to give a result something like this:

For this example, the i