OrionStar Wins Challenge to Recognize One Million Celebrity Faces with Artificial Intelligence

Researchers at MSR published the first MS-Celeb-1M dataset, comprising images of real-life celebrities, in June 2016 to encourage the development of advanced facial recognition technologies.

BEIJING, July 26, 2017 /PRNewswire/ -- Beijing OrionStar Technology Co., Ltd. ("OrionStar"), a China-based artificial intelligence company with which Cheetah Mobile Inc. (NYSE: CMCM) ("Cheetah Mobile") has entered into an investment agreement, recently ranked number one in the "MS-Celeb-1M" challenge, a competition to recognize and identify images of one million celebrities in a pre-set database. The challenge was organized by Microsoft Research (MSR) for a workshop at ICCV 2017, the world's premier international computer vision event.


Researchers at MSR published the first MS-Celeb-1M dataset, comprising images of real-life celebrities, in June 2016 to encourage the development of advanced facial recognition technologies. At the same time, they launched the first MS-Celeb-1M facial recognition challenge. Participants in the challenge are tasked with using a dataset provided by MSR as training data to develop an image recognition system capable of recognizing one million celebrities.

To evaluate the results, a subset of celebrities was selected from MSR's one-million celebrity list. Images of those celebrities were then carefully labelled and organized into two sets: a "random" set, in which one image is randomly selected for each celebrity from the labelled images; and a "hard" set, which includes the labelled image from each celebrity that is the most different from any other image in the training dataset.

Among this year's participants, OrionStar ranked first in the "Without External Data" category (only using the dataset provided) of the challenge with a 0.750 coverage probability for the random set and 0.606 for the hard set. During last year's challenge, MS-Celeb-1M @ ACM MM2016, the best performance without external data was 0.734 with the random set and 0.534 with the hard set (among 13 teams). The "Without External Data" challenge is significantly more difficult than the "With External Data" challenge, which allows competitors to access data from outside sources.

The facial recognition technology utilized by the OrionStar team in this challenge will have real-world applications for Cheetah Mobile products, including facial recognition software, robotics and mobile apps.

OrionStar's facial recognition technology is already being used in Cheetah Mobile's Live.me social live broadcasting app. Every day, Live.me produces more than 200,000 hours of live content. To ensure a wholesome experience, Live.me employs a 24-hour real-time monitoring system based on facial recognition technology developed by OrionStar. The technology has other uses as well, including the ability to recognize brand logos and support ad testing, generate user tags based on a variety of characteristics, and provide personalized content recommendations. It also supports all types of dynamic facial mapping.

About OrionStar

OrionStar is an artificial intelligence company incorporated in China and controlled by Mr. Sheng Fu, the chief executive officer and director of Cheetah Mobile. The company was established in September 2016 by a group of technology industry leaders and product specialists from Silicon Valley, Japan, Taiwan, Beijing and Shenzhen. In May 2017, Beijing Kingsoft Internet Security Software Co., Ltd. ("Beijing Security"), a wholly-owned subsidiary of Cheetah Mobile, entered into a capital injection agreement with OrionStar. Pursuant to the agreement, Beijing Security agreed to subscribe to newly issued equity interest in OrionStar.

OrionStar is committed to using AI technology to develop the next generation of ground-breaking technology products and free people from the burden of overly complicated tasks, make homes more intelligent and create a better world through technology.

In less than one year, OrionStar has successfully released its own voice recognition and image recognition systems.

In June 2017, OrionStar partnered with popular audio streaming platform Ximalaya.com in China to release Xiao Ya, a voice-operated speaker powered by OrionStar's self-developed remote voice interactive system, and the first product to connect to OrionStar's AI ecosystem.

About Cheetah Mobile

Cheetah Mobile is a leading mobile internet company. It aims to provide leading apps for mobile users worldwide and connect users with personalized content on the mobile platform.

Cheetah Mobile's products, including its popular utility applications Clean Master, CM Security and Battery Doctor, help make users' mobile internet experience smarter, speedier, and safer. The Company has attracted approximately 600 million global Mobile MAUs, of which approximately 80% are located outside of China. Leveraging the success of its utility applications, Cheetah Mobile has launched its line of mobile content-driven applications, including News Republic and Live.me.

Cheetah Mobile provides its advertising customers, which include direct advertisers and mobile advertising networks through which advertisers place their advertisements, with direct access to highly targeted mobile users and global promotional channels, which are capable of delivering targeted content to hundreds of millions of users.

Featured Product

ElectroCraft's Motion Control for Mobile Robots

ElectroCraft's Motion Control for Mobile Robots

ElectroCraft is showcasing its award-winning mobile robot technology including their powerful and compact wheel drives, high-torque-density brushless DC motors, precision linear actuators as well as servo motor drive technology at a variety of conferences and tradeshows including the Boston Robotics Summit. Robotics Summit is the premier symposium for the sharing of ideas, technology, and market developments for robotic technologies across industries. Beyond a showcase and pitch of product, ElectroCraft is eager to participate in the collaborative discussion of challenges and opportunities that will shape the near and long-term robotic marketplace.