How the future of "emotional recognition" will change education

How the future of "emotional recognition" will change education

Image source: American drama "Lie to me"

At the end of March, the future official WeChat public account used a push article to reveal its "face expression recognition" technology in the educational scene. A month later, Facethink, which focused on the development of the emotion recognition engine, announced that it had received a 10 million-level Pre-A round of financing from a good future.

A good future with the mission of “Promoting Education and Progress with Technology” has once again cast a technical “bomb” into the industry. How will this cutting-edge technology selected by the giant and the team behind it bring about education?

A "black technology" that is commercially viable

“Emotional recognition is not an emerging technology. It has been a few decades since Paul Ekman studied the relationship between expression and emotion.” FaceThink founder Yang Songfan said, “Now, the machine recognizes human expressions. It is already at a stage that is sufficient for commercial use."

Before deciding to start a business, Yang Songfan was already one of the most authoritative scholars in the field of domestic emotion recognition. When studying for doctoral and postdoctoral degrees at the University of California, Riverside, and Irvine, he took computer vision and emotional computing as his research direction. In the 2011 IEEE competition, Yang Songfan overcame 13 top research teams including MIT, Carnegie Mellon and Cambridge on behalf of the school in the field of emotional computing recognition.

After completing his studies, Yang Songfan joined Blizzard Entertainment, a well-known game developer, responsible for data mining and deep learning models. After returning to China in 2014, he served as an associate professor at Sichuan University and still specializes in emotional recognition. Talking about the “turning around” from research to entrepreneurship, Yang Songfan said: “The application of artificial intelligence technology needs to be combined with the actual situation of the scene. This requires a lot of data from the industry. So, to make a product that is actually applied may be A better choice."

The technology of “emotional recognition” itself has also come to the window from the laboratory to the industry. Yang Songfan told the mustard heap that since the deep neural network made breakthroughs in the field of image recognition in 2012, this underlying technology has begun to bear fruit in many fields. For example, in the field of expression recognition, it helps solve the problem of recognition under dark light and complex light, which makes it practically valuable.

In terms of application, world-renowned industry research companies such as Synovate and Ipsos have begun to apply this technology gradually, collecting emotional data while respondents participate in surveys and tests, and adding more reference dimensions to the survey results.

Technological advances and practical applications in individual industries have made Yang Songfan aware that the opportunity to commercialize “emotional recognition” has arrived. In 2016, Chen Fei, a Ph.D. in computer science at Tsinghua University, who brought his friends and classmates, began to form a team and founded FaceThink together.

Process information, how to make education "upgrade"?

In foreign countries, the technology of emotion recognition has been applied in different scenarios such as video research, pan-entertainment, movies, and amusement parks. In 2015, the emotional robot papper launched by Softbank Japan is based on emotional recognition and emotional feedback. However, Yang Songfan believes that education may be the most suitable industry for emotional outbreaks.

First, for "emotional recognition" itself, the difficulty of educational scenarios is lower than in other industries. Taking the scene of online teaching as an example, the image captured by the camera is usually relatively stable, and the face is the main body, which creates a good environment for the recognition of facial expressions by the machine. At the same time, the relatively standardized teaching process also allows student behaviors such as lectures, notes, and interactions to be estimated in advance, and the accuracy of recognition is greatly improved.

More critical is the actual demand in the education scene. Emotional recognition can be solved by the industry and is actually the "data" problem of the educational process. Experienced teachers can usually see the actual state of the students, but it is difficult to adjust for individual students, and can not form effective data after class. Even if it is taught online, the student's learning status data is often missing.

“In the past, we only evaluated students from the test scores,” Yang Songfan said. “And the process data actually adds a dimension of evaluation to education.”

How the future of "emotional recognition" will change education

When the coordinates change from one-dimensional to two-dimensional, the evaluation criteria are no longer limited to the results, but the comprehensive state of the class, analyzing whether the students are "not listening" or "not understood", thus providing targeted counseling to the students. If such analysis is refined to each knowledge point, it can be docked with the students' practice data to achieve a more granular and personalized education.

For schools, educational institutions and teachers themselves, the emotional data of students also has guiding significance for teaching and research. A large area of ​​negative emotional feedback may mean that the teaching arrangement is unreasonable. Improvements based on emotional data will provide students with a better classroom experience.

Provide solutions for the industry, products available within one year

As early as the financing announcement, the FaceThink team began to work closely with the future to polish the products. At present, the two sides are constantly testing and optimizing on the two scenarios of online teaching and double teacher classroom.

According to reports, in addition to common emotional indicators such as happiness, anger, surprise, fear, disgust, etc., FaceThink models the most important "concentration" in the educational scene, trying to make the results of machine recognition close to experienced teachers. According to the test, the recognition rate of FaceThink in the two scenarios is 94% and 91% respectively. As the amount of data increases, the recognition rate still has room for further improvement.

Another module developed specifically for educational scenarios is "line of sight recognition," which allows the machine to identify where the student is looking at the screen based on the screen given by the camera.

Yang Songfan said that FaceThink will eventually export the ability of emotion recognition to the industry in the form of To B. According to different partners, there are different ways of cooperation. The standardized SDK allows companies to add emotion recognition functions to their own products. For special scenarios, FaceThink can also provide customized products.

Yang Songfan is not worried about the acceptance of new technologies and new products by enterprises, teachers and students. He emphasized that the significance of emotion recognition is to improve classroom teaching and provide procedural data, rather than monitoring both teachers and students, which will ultimately benefit both parties. For institutional operators, emotional recognition is an effective tool to improve the quality of teaching. It can help them not only focus on operational data such as “renewance rate”, but also guide the industry to further development.

It is said that this emotional recognition product for the education industry will be the fastest in three quarters. After image and speech recognition, the “infiltration” of artificial intelligence technology to the teaching and research link is about to begin.

Electric Standing Desk

An electric adjustable-height Standing Desk is a desk typically powered by precision linear motor drives that easily changes height with the press of a button. They have a distinct advantage over most manually-operated desks in speed, convenience, and performance, but tend to cost more than their crank-operated cousins. Nevertheless, they remain the active standing workstation of choice for homes and offices across the country.

Electric Adjustable Desk

Electric Standing Desk,Electric Desk,Standing Desk Electric,Electric Adjustable Desk,Standing Office Desk

Suzhou Uplift Intelligent Technology Co., Ltd , https://www.uplifting-desk.com

Posted on