"This 'cognitive' sensor, which will be located next to the main image sensor, is designed to be used only for face recognition, object recognition, and QR code recognition. AI chip technology from UX Factory and technology from a domestic image sensor design company are combined. This chip is an ultra-low-power chip that reduces the amount of power when an electronic device recognizes an object by a hundredth of a conventional sensor.
“If the existing image sensor's object recognition operating power was 1 W (watt), the product developed this time can reduce the power to a maximum of 10 mW (milliwatt)”
The new image sensor appears to be a continuation of K-Eye cooperation project with KAIST presented at ISSCC in 2017.
"The goal of Park is to promote a sample chip in the second half, and then to mass-produce the chip in the first half of next year and apply it to home appliances."
Typo in the text I guess, should be 10 mW (milliwatt) probably.
ReplyDelete10 kW is quite a lot for an image sensor, even if that is the upper limit!
ReplyDeleteYes, the typo is in the source article. I've fixed it now.
ReplyDeleteany details on the image sensor being used? Is that a low resolution rolling shutter? Or is there some logic in the pixel level?
ReplyDelete