Due to recent advances in machine learning, there has been an explosive development of multiple methodologies that automatically extract information from architectural floor plans. Nevertheless, the lack of a standard notation and the high variability in style and composition make it urgent to devise reliable and effective approaches to analyze and recognize objects like walls, doors, and rooms from rasterized images. For such reason, and with the aim of bringing some significant contribution to the state-of-the-art, this paper provides a critical revision of the methodologies and tools from rule-based and learning-based approaches between the years 1995 to 2021. Datasets, scopes, and algorithms were discussed to guide future developers to improve productivity and reduce costs in the construction and design industries. This study concludes that most research relies on a particular plan style, facing problems regarding generalization and comparison due to the lack of a standard metric and the limited public datasets. However, the study also highlights that combining existing tasks can be employed in various and increasing applications.