{"id":11948,"date":"2018-04-24T06:50:36","date_gmt":"2018-04-24T06:50:36","guid":{"rendered":"http:\/\/www.lifeandnews.com\/articles\/?p=11948"},"modified":"2018-04-27T07:27:46","modified_gmt":"2018-04-27T07:27:46","slug":"i-run-facial-recognition-on-buildings-to-unlock-architectural-secrets","status":"publish","type":"post","link":"https:\/\/www.lifeandnews.com\/articles\/i-run-facial-recognition-on-buildings-to-unlock-architectural-secrets\/","title":{"rendered":"I run ‘facial recognition’ on buildings to unlock architectural secrets"},"content":{"rendered":"

Peter Christensen<\/a>, University of Rochester<\/a><\/em><\/span><\/p>\n

About a decade ago, a modest update to Apple\u2019s iPhoto software showed me a new way to study architectural history. The February 2009 update added facial recognition<\/a>, allowing users to tag friends and loved ones in their photos. After a few faces were tagged, the software would begin to offer suggestions. <\/p>\n

But it wasn\u2019t always accurate. Though Apple\u2019s algorithm continues to improve, it had a tendency to find faces in objects \u2013 not just statues or sculptures of people, but even cats or Christmas trees. For me, the possibilities became clearest when iPhoto confused a human friend of mine \u2013 I\u2019ll call him Mike \u2013 with a building called the Great Mosque of Cordoba<\/a>.<\/p>\n

\n \"\"<\/a>
\n People \u2013 but maybe not computers \u2013 can tell whether this is a person\u2019s face or the Great Mosque of Cordoba.<\/span>
\n Erinc Salor<\/a>, CC BY-SA<\/a><\/span>
\n <\/figcaption><\/figure>\n

The ceiling of the mosque\u2019s forecourt supposedly resembled Mike\u2019s brown hair. The layering of two Visigothic archways supposedly resembled the area between Mike\u2019s hairline and the edge of his brow. Finally, the related alignment of the Moorish cusped arches with their striped stonework resembled Mike\u2019s eyes and nose just enough that the software thought a 10th-century mosque was the face of a 21st-century human. <\/p>\n

Rather than viewing this as a failure, I realized I had found a new insight: Just as people\u2019s faces have features that can be recognized by algorithms, so do buildings. That began my effort to perform facial recognition on buildings \u2013 or, more formally, \u201carchitectural biometrics<\/a>.\u201d Buildings, like people, may just have biometric identities too.<\/p>\n

Facing the building<\/h2>\n

In the late 19th century, railway stations were built across Canada<\/a> and the Ottoman Empire<\/a>, as both countries sough to expand control of their territory and regional influence. In each country, a centralized team of architects was charged with designing dozens of similar-looking buildings to be constructed throughout a vast frontier landscape. Most of the designers had never been to the places their buildings would go, so they had no idea whether there were steep slopes, large rock outcroppings or other terrain variations that might have led to design changes.<\/p>\n

In both Canada and the Ottoman Empire, construction supervisors on the actual sites had to do their best to reconcile the official blueprints with what was possible on the ground. With communications slow and difficult, they often had to make their own changes to the buildings\u2019 designs to accommodate local topography, among other variable conditions.<\/p>\n

\n \"\"<\/a>
\n A composite image showing elements of train stations at Zeytinli, left, and Durak, right, which were built from the same plans, yet feature distinctive ornaments, windows and doors.<\/span>
\n Eitan Freedenberg<\/span>, CC BY-ND<\/a><\/span>
\n <\/figcaption><\/figure>\n

What\u2019s more, the people who actually did the building came from an ever-changing multinational labor force. In Canada, workers were Ukrainian, Chinese, Scandinavian and Native American; in the Ottoman Empire, workers were Arab, Greek and Kurdish. They had to follow directions given in languages they didn\u2019t speak and understand blueprints and drawings labeled in languages they didn\u2019t read. <\/p>\n

As a result, the engineers and workers\u2019 own cultural notions of what a building should look like and how it should be constructed left their figurative fingerprints on what was built, and how it looked. In each place, there are subtle differences. Some stations\u2019 wooden window frames are beveled, some roofs have finials, and some rounded arches are replaced with ever-so-slightly pointed arches.<\/p>\n

Other design changes may have happened more recently, with renovations and restorations. Meanwhile, time has worn down materials, weather has damaged structures and, in some cases, animals have added their own elements \u2013 like birds\u2019 nests. <\/p>\n

The people behind the facades<\/h2>\n

In the Canadian and Ottoman case studies, many people had opportunities to influence the final building. The variations are quite like differences between people\u2019s faces \u2013 most people have two eyes, a nose, a mouth and two ears, but exactly how those features are shaped and where they\u2019re placed can vary.<\/p>\n

Thinking about buildings as objects with biometric identities, I began to use analysis similar to facial recognition to find the subtle differences in each building. My team and I used laser scanners to take detailed 3-D measurements of railway stations in Turkey and Canada. We processed the raw data to create computerized models of those measurements. <\/p>\n

\n \"\"<\/a>
\n Digital scans of buildings lets researchers compare similarities and differences.<\/span>
\n Peter Christensen<\/span>, CC BY-ND<\/a><\/span>
\n <\/figcaption><\/figure>\n

That, in turn, revealed the hands of the builders, highlighting the geographic and multicultural influences<\/a> that shaped the resulting buildings.<\/p>\n

This evidence called into question previous assumptions that buildings, like a sculpture or a painting, are primarily influenced by just one person<\/a>. Our work has shown that buildings really only begin with drawings, but then invite the input of a vast number of creators, most of whom never achieve the heroic status of architect or designer.<\/p>\n

To date, there are no good methods to even try to identify these people and highlight their artistic choices. The absence of their voices has only tended to prop up the idea that architecture is made only by brilliant individuals. <\/p>\n

\"TheAs 3-D scanners become increasingly common<\/a>, perhaps even elements of smartphones<\/a>, our method will be available to almost anyone. People will use this technology on large objects like buildings, but small ones too. At present, our group is working with Paleoindian points, more commonly known as \u201carrowheads,\u201d to explore a very different history, geography and set of circumstances than we did with the railway stations.<\/p>\n

\n