Google Engineer Creates AI Tool That Identifies Anonymous Faces In WWII Photos

Photo 83082760 © Bratty1206 | Dreamstime.com

The present has access to a deluge of photos from the Holocaust, but many of the subjects in them remain unknown. 80 years on, one Google software engineer from New York and his growing team of engineers, data scientists, and researchers, are hoping to identify these “invisible” figures to bring closure to survivors and their families.

In his free time, Daniel Patt runs Numbers to Names (N2N), a facial-recognition artificial intelligence platform that scans hundreds of thousands of photos from pre-war Europe and the Holocast—dating back to 1914–1945—that hail from the collection of the United States Holocaust Memorial Museum (USHMM) to identify some of the people in those pictures.

N2N is free for the public to use, and visitors can simply upload a photo from these eras to potentially find a match with the database.

Photo 76336414 © Giuseppe Crimeni | Dreamstime.com

The web tool, however, makes no promises of perfectly identifying subjects. More specifically, it sends back 10 of the most likely matches from its dataset, along with similarity scores. It’s up to the user to ascertain whether any of the individuals in the results is a long-lost loved one, Patt tells The Times of Israel, which first profiled the project for Holocaust Survivor Day on June 26.

All told, N2N has successfully linked some anonymous faces to survivors and their descendants, who include celebrities, says the report. One of them is Canadian musician Geddy Lee, who affirmed that a woman in a photo taken at the displaced persons camp in Bergen-Belsen was indeed his mother.

Patt acknowledges that the cost to run the site will increase as more photos, and now videos, are being studied by the AI. To keep costs low in the future, he and his team are considering rolling out an offline search for users to find a match within a day, instead of seconds.

The software engineer, 40, was inspired to create N2N after a visit to the POLIN Museum of the History of Polish Jews in Warsaw. As the grandson of three Polish survivors of the Holocaust, he felt compelled to help his family and others discover photos of loved ones killed during the Shoah.

One of Patt’s grandmothers, who is 91, was just nine years old when she escaped her hometown of Zamosc with her father and siblings. Her mother stayed behind and was eventually shot and killed.

The engineer couldn’t shake off the feeling that he was possibly walking past the faces of murdered relatives at the museum, and built the search engine to bring comfort to others in the same boat.

The next step for Patt and his teammates is to gain access to 700,000 more photos from these eras. In time, he anticipates that more people, such as students, could help run the software and discover new connections.

He notes the urgency of identifying anonymous figures in these photos. It is, of course, hoped that connections can be made when the grieving survivors are still alive.

[via The Times of Israel and Mashable, images via various sources]