The results of various challenges at CVPR 2020, the world's top conference in the field of computer vision, were recently announced. The AI Engineering Institute of TAL Education Group stood out among 69 outstanding teams worldwide and won the championship in the EmotioNet facial expression recognition competition.

Winning the authoritative academic competition showcases China's AI competitiveness to the world

The CVPR (Conference on Computer Vision and Pattern Recognition), known as the "Olympics" of the computer vision field, has held in Seattle, USA in June 2020.

The EmotioNet competition, an important part of this conference, is one of the most authoritative international academic competitions in the field of facial expression recognition. Its research results have been published in CVPR and ICCV (International Conference on Computer Vision), the three major authoritative conferences in computer vision. Over the years, it has attracted teams from many top international universities and enterprises to participate.

EmotioNet 2020 was organized by CVPR and hosted by The Ohio State University. Many professional teams, including those from TAL Education Group, Alibaba, University of Science and Technology of China, Chinese Academy of Sciences, Otto von Guericke University Magdeburg, and Nanyang Technological University, participated in this competition.

After intense competition, the research team from TAL Education Group ultimately won the championship, demonstrating China's hard power in AI technology field to the world.

In the competition, the AI Engineering Institute of TAL Education Group proposed a solution to the problem of inconsistent overfitting in multi-task learning for facial expression recognition, significantly improving the model's accuracy and F-score. This is one of the many basic AI capabilities in TAL Education Group's educational scenarios.

Currently, this technology has been widely applied in TAL Education Group's teaching business and shared with the entire industry through TAL Education Group's AI Open Platform.

(TAL Education Group wins championship of the EmotioNet facial expression recognition competition)

To break the difficult situation of AI technology implementation, we need to continuously increase investment in scientific research.

For a long time, the difficulty in implementing AI technology has been a common problem in the industry. The main reason is that its application requires a large amount of data and rich scene support.

Over the past decade of development, TAL Education Group has accumulated a vast amount of data resources and continuously explored offline face-to-face teaching, online dual-teacher teaching, and online one-on-one teaching and other educational scenarios, deeply understanding the needs of educational users. This has enabled TAL Education Group to have certain advantages in exploring the integration of AI and education. It could form a closed loop from research and development to application, achieve rapid iteration of AI products, and provide solutions and resource experience to the industry.

TAL Education Group has always adhered to the mission of "Love and technology make education better", continuously increasing investment in scientific research. It set up research departments such as the AI Engineering Institute, Brain Science Laboratory, Silicon Valley Research and Development Center, Big Data Center, and Postdoctoral Research Workstation.

Thanks to this, TAL Education Group has continuously made breakthroughs in the research and development of core technologies in the forefront.

In February this year, at the international artificial intelligence academic conference AAAI-2020, TAL Education Group was invited to showcase two of its latest academic research results and lead the organization of an academic seminar on "Intelligent Education".

As of now, TAL Education Group has developed more than 100 AI capabilities including image, voice, data mining, and natural language processing, covering eight types, and has created more than ten AI solutions for educational scenarios, covering all teaching links such as teaching, learning, testing, practice, and evaluation. Many of these capabilities are unique or leading in the industry. At the same time, it has independently built and developed multiple AI platforms and infrastructure such as storage, annotation, training, and online services.

Taking on the responsibility of the "national team" to create a diversified new ecosystem of intelligent education

In recent years, the development of AI in China's education sector has been rapid. As a leading enterprise in the education industry, TAL Education Group is actively taking on the responsibility of promoting the integration of AI and education, and working with industry partners to promote the intelligent upgrade of the education industry.

In August 2019, the Ministry of Science and Technology approved the construction of the National New Generation Artificial Intelligence Open Innovation Platform for Smart Education based on TAL Education Group, marking that TAL Education Group has officially become the "national team" in the field of education + AI.

The good construction of the National New Generation Artificial Intelligence Open Innovation Platform for Smart Education will strongly support the deep integration of AI and education, help promote the empowerment of smart education to all education participants, effectively improve the teaching efficiency of teachers and the management level of schools, and solve the problem of unbalanced educational resources. Ultimately, it will achieve scalable individualized teaching through data-driven methods and promote the transformation of educational concepts and models. At the same time, it will lower the technical research and development threshold for small and medium-sized enterprises and reduce the application and transformation risks for educational institutions, stimulating the vitality of smart education innovation and entrepreneurship.

Since the approval of the platform's construction, TAL Education Group has had more than 100 AI capabilities available for industry opening. To promote the common progress of the industry, TAL Education Group has open-sourced five educational scene data sets, shared seven AI solutions, and provided platform services to the outside world. In addition, TAL Education Group has established technical cooperation with prestigious universities such as Tsinghua University, the Institute of Computing Technology of the Chinese Academy of Sciences, Beijing Normal University, and Stanford University. It has also joined hands with numerous public schools, educational training institutions, research institutions, and technology enterprises to build an open ecosystem.

TAL Education Group stated that it hopes to collaborate with all sectors to address the core demands and key issues of future smart education applications and the development of the education industry. By focusing on common technologies and platform services in the field of artificial intelligence throughout the entire education process and all scenarios, it aims to form a smart education artificial intelligence open innovation platform characterized by "autonomous control of basic software and hardware, open sharing of industry common technologies and data, effective empowerment of application scenarios, and a vibrant industrial innovation ecosystem". Ultimately, it aspires to create a diversified new ecosystem of smart education that is "symbiotic", "mutually generative", and "innovative" in the future.