How ai invigilation technology invigilates eyeball recognition

Invigilate eyeballs through deep learning behavior prediction technology.

We have developed a system that can automatically detect whether candidates cheat, such as peeking at mobile phones, sending answers to candidates next door, and judging from the gestures, bones and sight of the characters in the film. By using deep learning instead of cloud in the device, the system can cover the faces of unsuspecting candidates, taking into account the privacy of candidates and helping to detect cheating.

Because the system does not need to install large-scale equipment, it can automatically monitor a large range to assist examiners only by using computer terminal equipment and photographic lenses on the spot. Cheating detection AI can give consideration to privacy, better assist visual invigilators, and provide a fairer examination environment for all candidates.

The hardware acceleration of deep learning has slowed down, and the pulsating array has brought huge acceleration growth to the world by 20 17. We can't expect the computing power of 20 19 to be greatly improved.

NVidia's Turing core is only a little faster than Volta's core. Google's TPUv3 system now uses liquid cooling, which is more dense than previous products. I don't think 20 19 will have any big structural improvement, so don't increase it too much as in previous years.

Summarized as follows:

However, we will see that the new architectures of GraphCore and Gyrfalcon avoid the power consumption of memory transmission and support sparse computing, but the deep learning format needs to be changed to adapt to these new architectures, and new hardware research is needed, which is inspired by biological nano-intentionality.