Senior director of Baidu’s Institute of Deep Learning (IDL), Lin Yuanqing, introduced on Friday at the 2016 Chinese Association for Artificial Intelligence (CAAI) the six major directions of the IDL: deep learning platforms, video analysis, augmented reality (AR) technologies, basic image recognition, fine-grained recognition, and medical image recognition.
While Baidu’s Optical Character Recognition (OCR) technology allows smartphones to recognize an ID card with a precision rate of over 99%, Lin is confident that their AT technologies can start to make profits from this year.
“The market has an insatiable appetite for AI,” said Lin. “The point is to locate what the specific needs are and find out ways to meet them.”
In 2013, Baidu launched IDL, a team that is devoted to R&D of its own AI technologies. It covers areas including image recognition, machine learning, robotics, human-computer interaction, 3D vision, and heterogeneous computing. But it does not cover everything – for example, Baidu has abandoned drones, according to Chinese media The Paper.
Though admitting that the drone market is promising, without making drones by itself, Baidu found little market value remained to compete with leading drone maker DJI, said Lin. Nevertheless, relevant technologies can be applied to other areas.
Baidu’s unparalleled advantage in developing AI technologies lies in its great pool of data – the basis for deep learning. Companies without enough data may need to work with others, which also accounts for the acquisition of the AI company DeepMind by Google.
Lin said Baidu is also ready to acquire companies with potential that can complement Baidu’s picture.
“It takes a combined effort to boost the AI industry,” said Lin. At the 2016 Baidu World Conference on this Thursday, Baidu will open up some of its technologies to the public. For example, instead of working on OCR by themselves, companies can build upon Baidu’s world-leading OCR technology.
(Top photo from Baidu Images)