Invisible Mask: R&D for Privacy "Anti-face Recognition Technology" [Full Text]

Technology dynamics Unless you always wear a mask with you, it is almost impossible to completely avoid the lens and face recognition technology of OTHERS. This not only involves privacy issues, it also puts potential pressure on companies that provide customers with personal data protection services.

D-ID is a startup company and has just joined YCombinator, a well-known incubator in the United States. The meaning of D-ID stands for "deidentification." Want to solve this problem through the image processing tool, in this way, can interfere with face recognition technology in the picture looks and the original picture is almost the same.

Perry said: “When public officials in the security department are very concerned about face recognition technology, we begin to consider this issue. Now, this issue needs to attract everyone’s attention. Today’s streets are covered by cameras. Each of us We have always been photographed and we have been photographed. When we combine these cameras with face recognition technology, our privacy is completely out of the question." ATMs using face recognition technology have been deployed in Macau. It is coming, and the technology is also being tested at border control agencies in several countries. In China, face recognition technology is even used to identify people who do not obey traffic rules. So finding a way to deal with face recognition technology becomes even more urgent.

Developing “anti-face recognition technology” for privacy, letting everyone wear “invisible mask”

The growth of D-ID has also been driven by some data privacy regulations, such as the EU's general data protection regulations. These data privacy regulations will take effect in May 2018. It requires companies to conduct more stringent protection of personal data, including biometric data, or companies will face significant fines. D-ID stated that the technology it designed and developed is difficult to overcome by artificial intelligence. Perry declined to say how D-ID's algorithm specifically does this, but he stated that the company’s goal is to become a “standard for image protection” and to protect every online shared photo that contains biometric data .

D-ID mainly provides services in three vertical areas: one is the company that needs to provide protection for the images of employees and customers, the second is the health management organization, and the third is the government and security agencies that want to protect the biometric data. Perry said that the company will launch a pilot project, cloud-based image management service Cloudinary, which can protect more than 14 billion media assets.

Other companies are also developing similar technologies to protect image data from recognition by face recognition technologies, including those that specifically help organizations organize compliance with privacy laws or provide SaaS-based data protection. Most of them take technology that makes the face on the image completely unrecognizable, but Perry said that the D-ID is different because under the D-ID technology, the change of the image is impossible to detect, at least. It is invisible to the human eye. This means that D-ID technology can appeal to people who just want to protect the pictures they post on the web. If there is enough demand, the company may release an application.

“It’s very important right now. With the progress of deep learning, every corner is covered with cameras, and regulators are aware of this. Now, there are more competitors who have joined the competition in this area. This is a good thing. In order to effectively protect personal data and prevent face recognition technology from invading our privacy, we need to accelerate the pace of development. We hope we can have as much influence on this aspect as possible. It can provide more protection for people's privacy," Perry declared.