Use Case: Student Emotion Recognition
What’s the issue?
Can the performance of students and tutors be improved by a combination of emotion recognition and artificial intelligence (AI)?
A number of universities are already looking at to the possibilities of using video monitoring and webcams across the intelligent campus along with emotion recognition software. In lecture theatres and learning spaces, disengaged or struggling students, could be identified and feedback provided to their tutor or lecturer, possibly in real time. In libraries and learning resource centres systems might recognise confused or distressed students allowing appropriate action to be taken.
Emotion recognition systems are already being piloted in the marketing and security industries.
Marketing companies are using the emotion recognition technology to determine customer’s level of attention and engagement. These results can be used to develop and launch new products.
Security software developers are claiming that emotion recognition systems, using CCTV footage, can identify possible criminals before they commit a crime.
These developments might be adapted and used to improve the learning experience for students both on and off the campus.
It is even envisaged that AI techniques, along with emotion recognition, could lead to the “digital tutor” that tailors courses to each student’s requirements.
Are there any current examples?
The Sichuan University in China has been using facial recognition technology for attendance monitoring for some time and is now investigating emotion recognition. The aim is to determine the student’s interest level, identify sadness, happiness and boredom as well as nodding or turning of the head. This data can then influence teaching techniques and content to ensure that students are stimulated and paying attention.
In Paris the ESG business school is planning to use a combination of facial recognition and AI to monitor student attention levels. They are focusing on online courses using webcams to record eye movement as well as facial emotions. If a low level of engagement is identified the system will create quizzes in real time to test content delivered during the inattentive period.
What issues are there to be concerned about?
There will clearly be concerns regarding the misuse and security of the data being collected about students using emotion recognition. Those involved in developing systems are well aware of these concerns and endeavouring to address them by protecting data through encryption and only storing data while it is being used.
There are also real concerns about the reliability of the emotion recognition systems. Misinterpretation of expression, such as confusion or concentration, can be an issue for humans, let alone algorithms and AI systems. Individuals often express similar emotions differently, indeed there are cross cultural differences as well. This issue was explored in by Shioiri, Someya, Helmeste and Tang in their paper Misinterpretation of Facial Expression which looked at Japanese and Caucasian expression differences.
It is clear that the level of errors made will need to be very low before use of these techniques will be widely accepted and used.