AI manufacturers are running on these two roads to explore more general large language models upwards and downwards to find more specific application scenarios.
Among many application scenarios, education is one of the most important application directions of AI. China has 291 million students, 18.4437 million full-time teachers, and 28 million campus terminal devices. The huge number of users has gathered massive data resources. Compared with other industries, China's education is also highly digitized. Statistics show that the Internet access rate of primary and secondary schools is as high as 100%. This means that "AI+education" is fully charged.
Because of this, many technology companies choose to pre-emptively strike on "AI+education". Recently, Geling Shentong released three major smart campus sports products, namely Shentong Atongmu Sports Training and Examination System, Shentong Atongmu Somatosensory Interactive System, and Shentong Atongmu Sports Big Data Analysis System. These three products cover six major scenarios of sports teaching and research, physical examinations, after-school services, special classrooms, daily training, and cloud games.
Geling Shentong named the whole solution "Shentong Atongmu", and said that it will be committed to building a scientific sports scene and creating a smart campus sports solution. And the name "Shentong Atongmu" inevitably reminds people of "Astro Boy with Iron Arm". The latter is the image definition that appeared to promote the cutting-edge atomic energy technology at that time. From this point of view, the name "Shentong Atongmu" may also hide Gelingshentong's ambitions for future education and its current technology focus.
In the digital age, sports and AI are gradually merging
Physical education is becoming more and more important.
According to the results of the Eighth National Survey on Students' Physique and Health, the rate of excellent physical fitness of high school students and college students has only increased by 1.8% and 0.2% in the past five years. At the same time, problems such as myopia and obesity have not been fully alleviated. The key to solving these problems lies in the reform of physical education.
In recent years, "Youth Physical Education" has been included in the strategic goals of the "Outline for Building a Powerful Sports Country". In April 2022, the Ministry of Education issued the "Compulsory Education Curriculum Plan and Curriculum Standards (2022 Edition)", which clearly stipulates that the proportion of "sports and health" class hours should be kept at 10%-11%, higher than the 6%-8% of foreign languages , only after Chinese and Mathematics. The phenomenon of physical education classes being taken for no reason may be becoming a thing of the past.
Compared with the rapid switching of the policy environment, the changes in traditional physical education classes are slightly slower. Curriculum learning can be disassembled into a loop of training process and testing process. Compared with other disciplines, the training and testing of physical education are more complicated. After all, evaluating a set of moves is not as difficult as marking a set of test papers. In traditional physical education classrooms, teachers either use extremely keen observation skills to complete teaching, or test students through highly abstract sports goals such as time and distance. The result is often that the teacher puts in a lot of energy, but the students have little interest.
The seemingly "lose-lose" situation actually only needs one key - accurate assessment brought by artificial intelligence. The key to this key is that AI turns the collection of subjective experience to the analysis of objective data. Specifically, the Smart Campus solution records the entire movement process through technologies such as face recognition, human body posture recognition, and motion capture, and on this basis, uses AI for real-time calculation, analysis, and prediction, and ultimately provides students with A personalized training program. It can be said that the digitization of the objective world by AI makes accurate assessment truly possible.
With the empowerment of AI, physical education teachers' energy has been greatly released, and classroom efficiency has been further improved. In addition to efficiency, AI can also make physical education safer, because each student's physical fitness is different, so they may have different physical reactions to the same sports, and the rational use of emerging technologies such as AI is conducive to More accurately assess the physical fitness of students. Based on this, some schools have tried to use AI technology to assist teaching. According to a report by China Sports Daily, Yanglin Middle School in Hefei City, Anhui Province has tried to let students wear sports heart rate belts in class. If a student's heart rate is too high during exercise, artificial intelligence will notify teachers and parents as soon as possible.
Improving teaching efficiency and ensuring sports safety... Artificial intelligence is rapidly bringing traditional physical education into the era of digital intelligence.
These changes have also given AI manufacturers new scope for use. In May last year, NetEase Youdao announced to upgrade its "Smart Youdao" product matrix and add a new smart sports solution. In August, HKUST Xunfei released smart sports solutions at the 2022 new product launch conference, emphasizing the use of AI to improve the quality and fun of school sports. Earlier, listed companies Jiafa Education and Oriental Venus have also released AI products related to smart sports.
In addition to the "Shentong Atongmu" smart sports solution released by Geling Shentong this time, there are already many mature solutions on the market. However, do these solutions really hit the pain points of the campus scene? What kind of program can better meet the requirements of school sports?
In the campus scene, what kind of AI solutions are needed for school sports?
"AI+sports" was applied to professional athletes earlier. Artificial intelligence can accurately analyze athletes' movement techniques and put forward effective suggestions based on biomechanics. Therefore, there are more and more cases of using AI technology to improve athletes' abilities on the field.
However, professional athletes and school students have different athletic goals. The goal of professional athletes is to constantly challenge their physical limits, while the goal of school students is to improve their physical fitness. So in the context of campus education, what kind of smart sports solutions do schools need?
Taking Yunnan Province as an example, the test items of the 2023 physical education high school entrance examination will be adjusted from "men's 1000 meters" and "women's 800 meters" to "brisk walking or jogging". There is no timing or time limit in the exam, only counting. In other words, as long as students participate enough times, they can get full marks. Judging from policy adjustments, "getting students moving" is the goal of campus sports. It can be inferred from this that AI solutions that allow students to "move" are also more suitable for campus scenes.
Of course, this solution that allows students to "move" must come from "incentives" rather than "coercion". The so-called "incentive" refers to strengthening the interest of students in sports, so that they can spontaneously carry out sports and fitness, such as points, leaderboards, and upgrade systems, which are lacking in traditional physical education. Interactive methods can stimulate students' sports to a certain extent interest.
But even more important is the positive feedback that builds movement. In the concept of gamification education, the establishment of a positive feedback mechanism can effectively enhance the attractiveness of physical education. Specifically, positive feedback can be divided into two parts: "how are you doing now" and "what to do next", corresponding to the evaluation and suggestions in physical education. More accurate assessment and more effective suggestions will undoubtedly allow students to get more positive feedback, and thus quickly fall in love with physical education.
Based on this, AI manufacturers such as Geling Shentong will focus on accurate visual recognition, comprehensive data analysis and real-time solutions, but there are considerable differences in the focus of each solution.
NetEase Youdao's smart sports solution is based on the "ability map", which is a set of indicator systems and algorithm models for students' core literacy. With this set of algorithms, NetEase Youdao can generate personal exercise reports based on exercise data. Through this report, students can understand their own speed quality, flexibility quality, endurance quality and other ability maps.
If NetEase’s key word is “generation” in the presentation of smart sports solutions, then the key word of HKUST Xunfei is “connection”. The "3" in the "136 system" of iFLYTEK's smart sports solution refers to the trinity of the Internet of Things, the Internet of Things, and the Internet of Intelligence. Through wearable devices, smart playgrounds and other facilities and venues, iFLYTEK has opened up the links of learning, practice and competition, providing one-stop smart services for campus sports.
In contrast, the differentiated advantage of Geling Deep Pupil lies in the recognition technology. It is understood that Geling Shentong has self-developed a 3D human body pose estimation algorithm, which can overcome difficulties such as inaccurate collection of key points of the human body and improve accuracy.
In fact, in the current physical education courses, 2D visual recognition technology is rarely used, mainly because the 2D skeleton can correspond to multiple 3D skeletons, and it is difficult to accurately obtain the key points of the 3D human body. However, Geling Shentong breaks through technical difficulties and directly jumps from 2D to 3D, which can make more accurate judgments on sports postures, which reflects its leading technology. It is reported that the face recognition technology of Geling Shentong has won the championship of NIST's global FRVT test, and the key point detection technology of the human body is also among the best in PRCV2019 and ICCV.
It is worth mentioning that, in conjunction with the AI technology platform "Deep Tong Brain", Geling Deep Tong can also control the recognition speed within 200ms, and can provide real-time data feedback and training suggestions for teachers and students. On the whole, the solution of Geling Shentong may be more in line with the actual needs of campus sports for positive feedback.
Image source: Geling Shentong Intelligent Sports Management System
However, on the whole, major manufacturers are more inclined to start from their own inherent technical advantages to build campus smart sports application scenarios, but only partial technical advantages are obviously not enough.
With the further development of smart sports education and the further convergence of AI manufacturers' technologies, the user's terminal experience will become more and more important. How to maximize the application value of artificial intelligence in sports scenes has become a problem that relevant manufacturers need to continue to explore. In this regard, Gelingshentong believes that it can make efforts in "human-computer interaction" to enhance the immersion and richness of experience, which also defines a breakthrough direction for the industry in the future. What is certain is that the future vision of smart physical education is not a simple accumulation of technologies, but the reconstruction of core scene services.
By: Tim Ze Tyler
Source: Songuo Finance