Privacy advocates are up in arms after the announcement that Chinese researchers have developed a 500-megapixel digital camera, sharp enough to identify individual faces in a crowd of thousands.
Debuted at the recent International Industry Fair in Shanghai, the camera was developed by researchers at Shanghai’s Fudan University and the Institute of Optics, Fine Mechanics and Physics at Changchun’s Chinese Academy of Sciences.
Its resolution is high enough to capture a panoramic image of an entire stadium and distinguish every face in the image, according to reports that lauded its potential to assist in public security, crowd monitoring, and surveillance of military bases or national borders.
The camera dwarfs consumer-grade cameras and smartphone sensors – which this year jumped to 61MP and 100MB, respectively – but its early stage and large size make it more suitable for fixed installations, such as ultra-high resolution microscopes or space-bound telescopes.
The high-resolution surveillance state
File size increases dramatically with image resolution, making it a serious impediment to fast processing: one of the world’s largest photos, a 365,000MP photo of Mont Blanc, required 70,000 images taken over 35 hours and took 2 months to process.
The difference is that the 500MP camera promises near real-time processing thanks to a pair of custom-built chips.
Those chips enable the camera to record photos and video at the same resolution, feeding images and video into a cloud-based artificial intelligence (AI) service where individuals can be quickly identified against databases of citizens.
The camera’s massive resolution has worried privacy advocates who believe the technology will be yet another tool for expanding Chinese monitoring and enforcing the Chinese government’s controversial Social Credit System, which rewards and penalises citizens based on their social and anti-social activities.
Zhima Credit, a subsidiary of Huangzhou-based computing giant Alibaba Group, was one of several companies chosen by the Chinese government to trial a credit-scoring environment that would reward and punish users based on their social-media interactions.
It’s a real-world use case that taps into Chinese companies’ hard push into AI research – which has seen Alibaba Group building AI capabilities into its fast-growing cloud service, which is being upgraded with a third-generation X-Dragon Architecture designed to boost cloud performance by a factor of five.
Alibaba’s T-Head research unit recently announced an ‘AI inference’ processor called the Hanguang 800 that, the firm said, could analyse and categorise one billion images in five minutes – 12 times faster than using previous technologies.
This level of computing power would enable near real-time identification of 500MP images of people in large public scenes – providing source material for the increasingly pervasive Social Credit System and enabling monitoring by authorities.
Even Australia is getting in on the game, with Victoria recently uploading all of the state’s driver license photos to the coming National Driver License Facial Recognition Solution (NDLFRS).