(Yicai Global) Nov. 19 -- The core business of China's artificial intelligence is estimated at close to CNY100 billion (USD14.4 billion) this year. Its rapid future development is likely to vault that figure over CNY160 billion by 2020.

These were the words of Zhang Zijun, general manager of the AI research center under the country's industry and information technology ministry, at the China Hi-Tech Fair held from Nov. 14 to Nov. 18 in Shenzhen.

Investment in the AI sector reached CNY12 billion last year, while that in the first three quarters was more than the sum of last year and 2016.

AI financing usually involves billions of US dollars, and the sector is overvalued. Optimism over AI's future does startups no good, however. Sustained financing comes under great pressure when valuations are too high. 

Institutional investors will find it very difficult to fund a new round when a previous one has been overvalued, said Wu Guofeng, SDIC Fund Management Investment director at the China Hi-Tech Fair, which runs in Shenzhen from Nov. 14 to 18.

"Reasonable valuation is necessary in both primary and secondary markets. A healthy environment for investment is better for startups," Wu said. "AI financing is often Billions of US dollars at a minimum, so the entire investment environment is unhealthy."

More business value can only be created  by accelerating implementation of the AI technologies attracting red-hot investment. Computer vision, for example, in China has obtained CNY23 billion in investment this year, claiming over one-third of that in the country's AI sector.

Looking Good

Computer vision is an AI technology where liquidity is the smoothest, Wu thinks. CV made up 37 percent of China's AI market last year and ranked first with CNY8 billion attracted, per data the China Academy of Information and Communications Technology released in February.

CV is most widely used in security image analysis, the proportion of which was about 68 percent last year. It is now being used in ever more fields, such as autonomous driving and medical imaging, but is not yet widely used and only in preliminary stages.

Face recognition is expected to be a leading biometric technology and has great potential, believes Michael Deng, chief executive of the Fremont, California-based visual AI supplier ArcSoft, adding that the technology  is less mature in object and scene identification and vast volumes of high-quality and fine sorted application scenario data are needed to update algorithms for specific application scenarios.

As the number of intelligent terminal devices rises, captured pictures and videos will contain an ever greater deal of data information, and so computer vision will play a key role, Deng thinks.

The automotive industry is another application field ArcSoft regards as important. The number of internet-connected vehicles is expected to reach 160 million in 2020 and each one may have a dozen cameras and various vision-related applications to form control and monitoring systems to ensure driver security and convenience, research data show.

Vehicle-mounted vision will be the next force driving visual AI technology. Future manufacturing will have wide scope to apply visual AI technology, Deng predicted.

ArcSoft will open a visual AI platform and hopes to have 100,000 users within two years to hasten application and ecological formation, Deng said.

Editor: Ben Armour