You’ve probably seen or read about apps and services that aggregate photos to get multiple perspectives of a single place or event at a place. You may also be familiar with projects that rely on people donating a portion of their computer’s time for massively difficult problems like detecting an extra-terrestrial signal or protein folding. Researchers at Rice University have combined these concepts to provide a way to search photos stored locally on smartphones to find objects (including people) in collections of crowd sourced photos in near real-time.
Unlike the services we have today, these photos remain on the smartphone. The photos are not uploaded to a central server for processing. The search processing itself takes place on the smartphone too. This system, named Theia, requires an app to be installed on the smartphone. So, this is not something that would just go through all of your photos without your knowledge. While it definitely has privacy and personal security aspects that should be of concern, it also has the potential to visually find things for good outcomes (like a missing person).
The researchers have a prototype running on Android phones. This is not a production application or service at this time.
Search the world’s smartphone photos (I Programmer)