AI classifies Emirati homes with high accuracy
Summary and headline written by AI from the source article. How we work
Researchers in the United Arab Emirates developed a machine learning system to identify architectural styles of residential buildings.
They combined images of buildings with descriptions from experts, then used OpenAI’s CLIP model to create a unified set of data. The team then used UMAP to reduce the data’s complexity and K-Means to group similar styles together. After manually labeling these groupings, the researchers trained a Support Vector Machine to automatically classify new buildings.
This system achieved 98% accuracy when sorting buildings into eight different style categories. This performance exceeds that of other similar studies. The researchers believe their approach shows how artificial intelligence can help analyze architectural heritage in a way that is both efficient and understandable. The system could help people explore and understand the unique architectural styles of the UAE.
