CONNECT WITH US

Digitimes Research: DNN-architecture AI widely used in image recognition

Jessie Lin, DIGITIMES Research, Taipei 0

Artificial intelligence (AI) based on deep neural network (DNN) architecture has been widely used in image recognition and has large potential in autonomous driving, according to Digitimes Research.

At the Consumer Electronics Show 2017 in early January, Nvidia unveiled Xavier, a SoC equipped with 8-core ARM processor and GPU based on Nvidia's next-generation graphics architecture Volta, specifically for use in automotive machine learning systems, Digitimes Research indicated. Israel-based Mobileye NV and STMicroelectronics have been jointly developing EyeQ5, an image-recognizing SoC for use in self-driving.

Other image-recognizing chip solutions are based on IP core or FPGAs. For example, Cadence Design System, Synopsis and CEVA have offered IP core-based products.

Due to complicated road conditions, self-driving entails various sensors to detect different conditions. DNN-based AI can precisely detect and classify conditions surrounding moving vehicles and combine various sensor data.

Content from this article was part of a complete Digitimes Research Chinese-language report that has not yet been translated into English. If you are interested in an English version of the report or wish to receive more information about the report, click here to contact us and we will get back to you as soon as possible. Digitimes Research also provides quarterly tracking services for market sectors such as China Smartphone, China Smartphone AP, Taiwan ICT and Taiwan FPD. Click here for more information about Digitimes Research Tracking services.