Does this repo support evaluating performance on 'my own dataset' (finding Precision and Recall for instance)? If so, should there be a little documentation on how to?

[MjdMahasneh]