$2.6 million award for computer vision in archaeology

A Brown University archaeologist and team of engineers have been awarded $2.6 million from the National Science Foundation to use computer vision and pattern recognition in an archaeological excavation. The team is setting out to change the way archaeologists conduct fieldwork by developing innovative techniques for excavation, reconstruction, and interpretation during the next four years.

The work will focus on the site of Apollonia-Arsuf, located on the Mediterranean coast in Israel. The site has been under archaeological excavation and conservation since the 1950s and was formally recognized in 2004 as one of the 100 most endangered world monuments by the World Monuments Fund.

The team's main goal is to develop a visual archaeological database (VAD), which will transform the current slow and tedious documentation process of an excavation. Video cameras and digital scanning stations will be installed around the site to maintain a continuous visual recording of excavation activity. Excavators will also be able to immediately process artifacts by inputting text, video, still images, dense-data 3-D laser scans, and spatial position coordinate information into the database. The VAD will allow for immediate and efficient access, search, and image and 3-D scene reconstruction of the collected data.

Archaeologist Katharina Galor, one of the principal investigators, says her field's current protocol — relying on written field notes — is an outdated system. "We use written words to describe our findings when we should be using images and video to make the documentation process more efficient, accurate, and accessible to others," she said. "The more precisely we can physically reproduce what is left, and the more accurately we can piece together broken and dispersed pieces, the greater our chances of reconstructing and understanding the past truthfully."