Google Research algorithm that removes image watermarks

Four Google researchers have uncovered a vulnerability in the way stock photography websites apply watermarks on their image portfolios.

The Google Research team exploited the vulnerability by creating an algorithm that removes watermarks from a site's images.

The algorithm achieved this after analyzing a portal's photos, detecting the repetitive watermark, and creating a mask that can be used to remove the semi-transparent layer from photos, restoring the original photo.

The algorithm works because all stock photo websites apply their watermark in the same way for all their image portfolio.

Google warned stock photography websites

Researchers said they carried out this research for good reasons, to improve the security of these portals and protect the copyright of users selling images through those websites.

Google staff privately disclosed their findings to all affected portals, along with countermeasures. Researchers recommended that stock photography portals apply watermarks with small variations on their photos.

Small changes to the watermark's shape, position, or transparency is enough to foil the algorithm they created, and prevent it from reconstructing the original image.

The Google Research algorithm is detailed in a paper titled "On The Effectiveness Of Visible Watermarks" that was recently presented at the 2017 Computer Vision and Pattern Recognition Conference (CVPR 2017). A video summary of the team's research is embedded below.

Related Articles:

Google Develops AI That Can Separate Voices in a Crowd

One In Every 200 Google Search Suggestions Is Polluted

Nvidia Develops AI That Reconstructs Corrupted Images

Google Accused of Showing ‘Total Contempt’ for Android Users’ Privacy

Russia Bans 1.8 Million Amazon and Google IPs in Attempt to Block Telegram