I would describe pl2sql as an ORM, as the name implies. However, I would describe Class::DBI and the others as an ROM, since it maps database functions (normally implemented in SQL) to class methods. Hence, Class::DBI defines numerous methods to correspond to their database equivalents. For nomenclature, this ORM designation seems somewhat Orwellian- its name has an opposite meaning. In fact, a Class::DBI object is exactly an RDB table represented as a Perl object.
PL2SQL takes the opposite approach. It provides automatic persistence (in an RDB table) of any arbitrary complex object. PL2SQL has two methods: The first assigns the object to a designated table via a unique ID. The second retrieves the object. It also uses a "Universal table definition" which provides reusability. Finally, a single table can be applied to numerous object types, instead of the one-to-one correspondence required of Class::DBI.
In fact, the "Universal table definition" is based on an XML representation of an arbitrary object. I lack the expertise to precisely place this module in the Universal jingo taxonomy, but based on Wikipedia, I would describe it as an XML-enabled XML database.