Person of Interest is an American television science fiction series that uses an Artificial Intelligence (AI) system to monitor, analyze, identify, and predict violent crimes.
By Zhu Long
CEO, Yitu Technology
Developed by billionaire genius Harold Finch, the AI operates cameras to track criminals and anticipate their actions. In addition, the system collaborates with other machines to rescue victims more quickly than humans ever could.
In this case, the often fantastic science fiction of television and film may not be so far-fetched. Facebook’s ‘Tag Suggestion’ software automatically scans new images seeking matches with previously tagged photos. The ‘FaceMe’ smart-face recognition algorithm from Taiwan’s CyberLink Corporation may be used on mobile devices to enable people-to-people connections by matching faces with personal information available across social media networks such as Facebook, Twitter, or LinkedIn.
Up to World Standards
The field of ‘computer vision’ began at the Massachusetts Institute of Technology (MIT) in 1966 when Marvin Minsky asked undergraduate student Gerald Jay Sussman to “spend the summer linking a camera to a computer and getting the computer to describe what it saw.” This innocent beginning led to the 1999 discovery of the Scale Invariant Feature Transform (SIFT), which provided precise comparisons between the same object in different images.
Since then, MIT has been joined by the University of California Berkeley and England’s Oxford University as the world’s top research institutions specializing in the field of computer vision. Talented people from these universities have either been lured away by giant companies like Google or have started their own businesses.
China began computer vision research in the 1990s. Performance improvements in Big Data computing and the seamless deployment of cameras have created the software and hardware infrastructure necessary to support China’s further development in this area. With the addition of AI, every move a person makes can be seen and future actions anticipated. With numerous Chinese startup companies in the area of computer vision, China has joined the leaders within this market segment across the globe.
Though not treated with the same public enthusiasm that was lavished on Google’s AlphaGo project, the combination of computer vision and AI is widely applied in public safety, finance, and information security.
Some experts feel it is possible that advanced computer vision technologies will change lives in ways that are quite different from what the academics and philosophers have so far forecast. Already, the AlphaGo has demonstrated that the gap between machine learning and human IQ is shrinking.
Expanding Human Capacities
Yitu Technology was founded in 2012 following my post-doctorate at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). I returned to China at that time to work with my middle school classmate Lin Chenxi, who was then project leader for Alibaba’s Aliyun Aspara Cloud Platform. Our new company combined my skills in computer vision with his expertise in cloud computing and Big Data technology.
Yitu’s first implementation was a video-image-based vehicle identification machine deployed in Suzhou, China. Right away, Yitu helped solve a burglary case involving the equivalent value of USD 15,000. After the break-in, a surveillance camera captured the suspect’s car driving away. Only 10 minutes following the incident report, the police detained the suspect by filtering car brands using the new vehicle identification system.
Face recognition will have a wider impact than vehicle identification. In a second case, the public security department in Suzhou identified 17 suspects out of 25 candidates using Yitu’s face recognition system, which produced early warnings that were generated by comparing a 100-million-image national database of criminals with images of Suzhou’s 13 million residents.
Vehicle and facial recognition technologies are also widely applied in the finance sector. For example, China Merchants Bank employs facial recognition technology in the Virtual Teller Machines (VTMs) of its 1,500 branches to open and activate accounts and perform other procedures that require identity verification. During testing, the technology proved accurate up to 98 percent of the time, while the rate of misrecognition was only 0.001 percent. The human eye can do no better than an error rate of 1 percent. In other words, machines have a far greater capacity for recognition accuracy than human beings.
Yitu Teams with Huawei
After years of successful practice, Yitu appreciates the opportunity to collaborate with Huawei on the construction of advanced Safe City solutions. We at Yitu are combining our deep understanding of computer vision technology with Huawei’s decades of ICT experience. Huawei’s open ICT ecosystem is a big help for startups like Yitu that want to quickly work out reliable solutions for users. In the future, we will be able to offer services to users in more countries and build safer cities using Huawei’s global network.
Source : ICT Insights