New Google research shows that a lot of stock watermarks, including those used by major websites, can be easily removed automatically by computers.
Google’s exploit, analysed hundreds of pictures with consistent watermarks, like Shutterstock’s. Once the algorithm learned to look at a photo and decide which pixel was a watermark and which wasn’t, it could just remove all the watermark pixels. Many watermarks are semi-transparent, so the algorithm already knew what should be in the watermark’s place as well.
“In ‘On The Effectiveness Of Visible Watermarks’ recently presented at the 2017 Computer Vision and Pattern Recognition Conference (CVPR 2017), we show that a computer algorithm can get past this protection and remove watermarks automatically, giving users unobstructed access to the clean images the watermarks are intended to protect,” Google said in its official blog post.
Shutterstock, however, seems to have found a hack to combat image stealing by developing a new randomised watermark.
Google’s research in the paper has said that many common stock watermarks can be removed since they appear identically across a huge number of online photos.
Interestingly, Google stated that applying a small spatial deformation to the watermarks during embedding can significantly degrade the quality of the watermark-removed images, with only imperceptible changes to the watermark itself.
“We found that introducing random geometric perturbations to the watermark — warping it when embedding it in each image — improves its robustness. Interestingly, very subtle warping is already enough to generate watermarks that this technique cannot fully defeat.
Engineers at Shutterstock have developed a new watermark randomizer that results in no two watermarks ever being exactly the same now.
But smaller outfits still have the watermark problem. Independent photographers often use software like Photoshop to type their names on images as a watermark. Other stock services might not have the engineering capabilities to move so quickly. If malicious AI developers take advantage of Google’s research, that could start a cat-and-mouse game for even Shutterstock.
Try deep learning using MATLAB