Shutterstock has reverse engineered Google’s watermark-removal algorithm

Stock photo vendors might not be all that screwed up in the end. Only a week after Google released a paper detailing how its researchers built an algorithm that automatically removes watermarks from stock photos, Shutterstock has already put together an antidote.

Taking a cue from the internet giant’s tips on how to strengthen watermarks, the popular stock photo distributor managed to reverse engineer Google’s software in order to stop copyright thieves from editing out watermarks and using their images for free.

To pull this off, its engineers built a smart watermark technology that counteracts the algorithm by deliberately inserting minor inconsistencies in the watermark patterns. The solution purportedly uses machine learning to continuously confuse Google’s software.

“The challenge was protecting images without degrading the image quality,” said Shutterstock CTO Martin Brodbeck. “Changing the opacity and location of a watermark does not make it more secure, however changing the geometry does.”

The solution came not without a little help from the source itself.

“Google published a white paper [PDF] outlining a way of using computer vision technology to eradicate watermarks from stock image collections on a large-scale,” Brodbeck added. “Shutterstock was notified before the paper was published and quickly began working to address the areas highlighted.”

Thanks to this collaboration between the two companies, Shutterstock’s new technology introduced several variables to its watermarks structures to make it difficult for programs to identify recurring patterns.

“The result was a watermark randomizer that our engineering team developed so that no two watermarks are the same,” Brodbeck explains. “The shapes vary per image and include contributor names. By creating a completely different watermark for each image, it makes it hard to truly identify the shape.”

According to the Shutterstock CTO, Google has already tested the new solution and found it to be functional. You can head to this page to see how the updated watermark protector works in action.