Where to instantiate objects stored in database?

Figuring out how to link object oriented code and database storage.

The Python MongoDB driver, PyMongo, returns results as dictionaries. I need to turn these into objects for further processing.

Assuming Data Access Objects (DAO) for each collection, hiding the MongoDB queries from the rest of the code.

Where should I create objects from their description in the database?

1. Create objects in main code

   The DAO method get_buildings_by_date() returns a building list as dictionaries.

   The total_area_by_date() method in the main code calls get_buildings_by_date() to get a list of building descriptions as dictionaries, instantiates Building, then does the computations using the methods in Building.

2. Create objects in DAO

   The DAO method get_buildings_by_age() instantiates Building and returns a list of Building instances.

   The total_area_by_date() method calls get_buildings_by_date(), then does the computations using the methods in Building.

3. Call DAO from objects themselves

   The calls to DAO are in the Building class itself, hidden from the total_area_by_date() method.

   Building class has static methods to return Building instances. For instance, it has a get_buildings_by_date() class method that calls the get_buildings_by_date() method from the DB manager layer, creates required number of Building instances and returns them as a list.

   total_area_by_date() calls Building.get_buildings_by_date(), then works with Building instances.

I think I'm missing a design pattern, here. Any idea which solution (or any other) I should pick?

I've read a bit about ODMs. Whether we will be using pymongo or an ODM is not decided, but tips about how ODMs in general or an ODM in particular answers/hides this issue are welcome.