Comment by JimDabell

Depending upon your use-case, you can get pretty good results by using spaCy for named entity recognition then matching on the titles of Wikipedia articles that have coördinates.