HomePageNewsCompany Announcement

Megvii open sources proprietary deep learning framework MegEngine

2020-03-26

Beijing, March 25, 2020 – Megvii Technology Limited announced today at an online press conference that it would open source its deep learning framework, MegEngine, to allow global developers to create AI solutions for industrial and commercial scenarios and to foster an ecosystem around its new generation AI productivity platform, Brain++.  MegEngine is being branded 天元(Tianyuan) in Chinese, a name which means “the origin for everything,” and also denotes the center point on the Go game board.

 

Initially developed in 2014 and now in its eighth edition, MegEngine is one of three primary components of the Brain++ architecture that Megvii built from scratch to train computer vision algorithms at scale. The other two proprietary components of Brain++ include MegData (a data management system) and MegCompute (a computing power dispatching system).

 

Megvii is one of the very few companies in the world that has developed its own deep learning framework, and by making parts of it available for free in this “alpha” launch, Megvii hopes that external developers both in and outside China, can make MegEngine even better for the benefit of all who use it. A beta launch is tentatively scheduled for June, and an official launch in September 2020.

 

YIN Qi, the co-founder, Chairman and CEO of Megvii said, My founders and I studied AI algorithms at Tsinghua in the late 2000s just after international academics achieved a major breakthrough in deep learning and openly shared their ideas.  Their work accelerated the fields of computer vision, speech recognition, natural language processing and audio recognition globally. In this same brave spirit, we want to open source some of our innovations, creating a dynamic where other engineers improve our tools, while we also have further influence on how the code evolves.


MegEngine will be released under the terms of Apache License 2.0, a permissive free software license written by the Apache Software Foundation that allows users to use the software for any purpose, to distribute it, to modify it, and to distribute modified versions of the software, without concern for royalties.  MegEngine code will be available on two code repository websites, GitHub and OpenI.  Tools and tutorials are available (in Chinese language only for now) at megengine.org.cn.

 


Co-founder and CTO TANG Wenbin said, “China has been participating in open source since the Android system was introduced in 2011, and Chinese developers today make up an increasingly large proportion of the global open source developer community.  This promotes international norms of intellectual property and coding.  We think that open sourcing MegEngine is consistent with this trend. Megvii wants to lead the way in China and create good will among developers here and abroad.

 

We have designed MegEngine to be a unified training and inference platform, with both static and dynamic graphs.  It’s fast and accurate, appeals to many types of developers, is optimized for many hardware and chip platforms, and is compatible with PyTorch models.  While we know aspects of MegEngine can be improved, we are also proud to contribute an elegantly designed, industrial strength system to existing global AI infrastructure.”

 

MegEngine has been deployed across Megvii’s three business units (personal IoT, city IoT, and supply chain IoT) and is used by its 1,400+ strong R&D team in its day-to-day algorithm and product development. Optimized for computer vision tasks, MegEngine is especially effective for training large volumes of images and videos and completing complex tasks such as image classification, object detection, object/scene segmentation and video analytics. It has helped Megvii win 27 first-place recognitions in some of the world’s most prestigious international AI challenges, including three first-place awards at the 2019 International Conference on Computer Vision COCO challenge. 

 

Megvii, supported by MegEngine, has developed advanced deep neural networks that are deployed across cloud, mobile and edge computing platforms. Notably, Megvii has developed ShuffleNet, a deep neural network designed for devices with relatively limited computing power such as mobile devices. According to tests conducted by Alphabet’s AI team in 2018, ShuffleNet achieved better performance than Alphabet’s comparable MobileNet in terms of both efficiency and algorithmic accuracy.

 

Chief Scientist and Head of Megvii Research Dr. SUN Jian said, “It took us six years from developing MegEngine for our own use in 2014 to open source it today. Open sourcing MegEngine is just the first step in our contributions to the open source movement.  We have used it to power our computational photography, facial recognition, and object recognition applications, but the developer community can use our foundational technology in innovative, real world applications of AI that we have not yet imagined. They will make what we built even better, and we can spend more time commercializing our products. It’s a win-win for all.”

 

At today’s online press conference, Megvii also announced that it will be introducing an enterprise version of Brain++ to offer support, performance, and managed AI productivity services for a fee to corporate clients. In addition, Megvii is releasing version two of its object detection database, www.objects365.org, which now has a dataset that is many times enlarged. 


 


Today’s conference featured keynote speeches from TANG and SUN as well as Dr. TIAN Zhongbo, Senior Technical Director of Megvii Research, and ZHAO Liwei, Vice President of Megvii’s Cloud Services. Hundreds of developers, academics and business partners attended the event.  Congratulatory messages and comments were received by Megvii from prominent figures including Tsinghua Professor Andrew Chi-Chih Yao and former Microsoft Executive Vice President of Artificial Intelligence & Research Harry Shum.

 

About Megvii Technology Limited

 

Megvii is a world-class, pure AI company with core competencies in deep learning. Founded in Beijing in 2011 by Yin Qi, Tang Wenbin and Yang Mu, three Tsinghua University graduates who studied under the Turing-prize winning Chinese computer scientist and computational theorist Andrew Chi-Chih Yao, Megvii is a pioneer in applying AI technology and computer vision algorithms to Internet of Things (IoT) use cases. The Company’s mission is to power humanity with AI. 

 

Megvii’s first commercialized AI product was a facial recognition solution. The Company has developed its new generation AI productivity platform Brain++, which includes a proprietary deep learning framework that functions as a unified underlying architecture and provides critical support tailored for the Company’s algorithm training and model improvement processes. 

 

Leveraging its strong software-hardware integration capabilities, Megvii offers full stack solutions to its customers, and has achieved leading positions in several key vertical markets, including personal IoT, city IoT and supply chain IoT. Megvii is building an AI infrastructure to connect and empower various IoT devices, to solve real life problems and pain points encountered by customers in sectors including fintech, transportation, logistics and others. Megvii is dedicated to helping customers achieve specific goals such as user authentication and fraud reduction, traffic optimization, and labor efficiency gains. 

 

Headquartered in Beijing with over 2,300 employees, Megvii maintains four research and development centers in China.

 

For press queries, please contact:

 

  Megvii Technology Limited

  Xie Yinan / Yang Xiaolan
 
 media@megvii.com

  +86-186-0067-8861

  +86-139-1027-0953

Brunswick Group
megvii@brunswickgroup.com

Queries in Chinese:

Yan Jiang / Amy Wang 

+852 9628-5168/ +86-132-6958-5046

 

Queries in English:

Sunitha Chalam / Ginny Wilmerding

+852 9628-3550 / +852-9730-6376


Megvii open sources proprietary deep learning framework MegEngine
2020-03-26

Beijing, March 25, 2020 – Megvii Technology Limited announced today at an online press conference that it would open source its deep learning framework, MegEngine, to allow global developers to create AI solutions for industrial and commercial scenarios and to foster an ecosystem around its new generation AI productivity platform, Brain++.  MegEngine is being branded 天元(Tianyuan) in Chinese, a name which means “the origin for everything,” and also denotes the center point on the Go game board.

 

Initially developed in 2014 and now in its eighth edition, MegEngine is one of three primary components of the Brain++ architecture that Megvii built from scratch to train computer vision algorithms at scale. The other two proprietary components of Brain++ include MegData (a data management system) and MegCompute (a computing power dispatching system).

 

Megvii is one of the very few companies in the world that has developed its own deep learning framework, and by making parts of it available for free in this “alpha” launch, Megvii hopes that external developers both in and outside China, can make MegEngine even better for the benefit of all who use it. A beta launch is tentatively scheduled for June, and an official launch in September 2020.

 

YIN Qi, the co-founder, Chairman and CEO of Megvii said, My founders and I studied AI algorithms at Tsinghua in the late 2000s just after international academics achieved a major breakthrough in deep learning and openly shared their ideas.  Their work accelerated the fields of computer vision, speech recognition, natural language processing and audio recognition globally. In this same brave spirit, we want to open source some of our innovations, creating a dynamic where other engineers improve our tools, while we also have further influence on how the code evolves.


MegEngine will be released under the terms of Apache License 2.0, a permissive free software license written by the Apache Software Foundation that allows users to use the software for any purpose, to distribute it, to modify it, and to distribute modified versions of the software, without concern for royalties.  MegEngine code will be available on two code repository websites, GitHub and OpenI.  Tools and tutorials are available (in Chinese language only for now) at megengine.org.cn.

 


Co-founder and CTO TANG Wenbin said, “China has been participating in open source since the Android system was introduced in 2011, and Chinese developers today make up an increasingly large proportion of the global open source developer community.  This promotes international norms of intellectual property and coding.  We think that open sourcing MegEngine is consistent with this trend. Megvii wants to lead the way in China and create good will among developers here and abroad.

 

We have designed MegEngine to be a unified training and inference platform, with both static and dynamic graphs.  It’s fast and accurate, appeals to many types of developers, is optimized for many hardware and chip platforms, and is compatible with PyTorch models.  While we know aspects of MegEngine can be improved, we are also proud to contribute an elegantly designed, industrial strength system to existing global AI infrastructure.”

 

MegEngine has been deployed across Megvii’s three business units (personal IoT, city IoT, and supply chain IoT) and is used by its 1,400+ strong R&D team in its day-to-day algorithm and product development. Optimized for computer vision tasks, MegEngine is especially effective for training large volumes of images and videos and completing complex tasks such as image classification, object detection, object/scene segmentation and video analytics. It has helped Megvii win 27 first-place recognitions in some of the world’s most prestigious international AI challenges, including three first-place awards at the 2019 International Conference on Computer Vision COCO challenge. 

 

Megvii, supported by MegEngine, has developed advanced deep neural networks that are deployed across cloud, mobile and edge computing platforms. Notably, Megvii has developed ShuffleNet, a deep neural network designed for devices with relatively limited computing power such as mobile devices. According to tests conducted by Alphabet’s AI team in 2018, ShuffleNet achieved better performance than Alphabet’s comparable MobileNet in terms of both efficiency and algorithmic accuracy.

 

Chief Scientist and Head of Megvii Research Dr. SUN Jian said, “It took us six years from developing MegEngine for our own use in 2014 to open source it today. Open sourcing MegEngine is just the first step in our contributions to the open source movement.  We have used it to power our computational photography, facial recognition, and object recognition applications, but the developer community can use our foundational technology in innovative, real world applications of AI that we have not yet imagined. They will make what we built even better, and we can spend more time commercializing our products. It’s a win-win for all.”

 

At today’s online press conference, Megvii also announced that it will be introducing an enterprise version of Brain++ to offer support, performance, and managed AI productivity services for a fee to corporate clients. In addition, Megvii is releasing version two of its object detection database, www.objects365.org, which now has a dataset that is many times enlarged. 


 


Today’s conference featured keynote speeches from TANG and SUN as well as Dr. TIAN Zhongbo, Senior Technical Director of Megvii Research, and ZHAO Liwei, Vice President of Megvii’s Cloud Services. Hundreds of developers, academics and business partners attended the event.  Congratulatory messages and comments were received by Megvii from prominent figures including Tsinghua Professor Andrew Chi-Chih Yao and former Microsoft Executive Vice President of Artificial Intelligence & Research Harry Shum.

 

About Megvii Technology Limited

 

Megvii is a world-class, pure AI company with core competencies in deep learning. Founded in Beijing in 2011 by Yin Qi, Tang Wenbin and Yang Mu, three Tsinghua University graduates who studied under the Turing-prize winning Chinese computer scientist and computational theorist Andrew Chi-Chih Yao, Megvii is a pioneer in applying AI technology and computer vision algorithms to Internet of Things (IoT) use cases. The Company’s mission is to power humanity with AI. 

 

Megvii’s first commercialized AI product was a facial recognition solution. The Company has developed its new generation AI productivity platform Brain++, which includes a proprietary deep learning framework that functions as a unified underlying architecture and provides critical support tailored for the Company’s algorithm training and model improvement processes. 

 

Leveraging its strong software-hardware integration capabilities, Megvii offers full stack solutions to its customers, and has achieved leading positions in several key vertical markets, including personal IoT, city IoT and supply chain IoT. Megvii is building an AI infrastructure to connect and empower various IoT devices, to solve real life problems and pain points encountered by customers in sectors including fintech, transportation, logistics and others. Megvii is dedicated to helping customers achieve specific goals such as user authentication and fraud reduction, traffic optimization, and labor efficiency gains. 

 

Headquartered in Beijing with over 2,300 employees, Megvii maintains four research and development centers in China.

 

For press queries, please contact:

 

  Megvii Technology Limited

  Xie Yinan / Yang Xiaolan
 
 media@megvii.com

  +86-186-0067-8861

  +86-139-1027-0953

Brunswick Group
megvii@brunswickgroup.com

Queries in Chinese:

Yan Jiang / Amy Wang 

+852 9628-5168/ +86-132-6958-5046

 

Queries in English:

Sunitha Chalam / Ginny Wilmerding

+852 9628-3550 / +852-9730-6376


Pre Sales Consultation
Leave us a message, we will feedback ASAP