The How’s and Why’s of Facial Recognition Technology in Education
This article explains the benefits and challenges of using facial recognition in education in real-life cases.
Facial recognition technology is making its way into more and more industries. According to Markets and Markets’ estimate, the facial recognition market size will grow from $3.8B in 2020 to $8.5B by 2025. The use of this technology by law enforcement agencies is a hot topic. However, the suggestion to use facial recognition in education may still puzzle some people.
But it shouldn’t. This technology can be instrumental in spotting criminals though it can be used for security purposes at schools and universities. The implementation of facial recognition in education is quite feasible, as it can make the teacher’s work easier and more effective. At the same time, the process of learning becomes more personalized and exciting.
How can the new technology make it happen? How does it work? What are the challenges on the way to using facial recognition in the education system? Let’s find out using Menklab experience implementing facial recognition technology for an ed-tech startup!
How Does Facial Recognition Work?
Before discussing the implementation of facial recognition technology in education, let’s find out more about these algorithms in general. Firstly, we will start with a general outline. Secondly, we will briefly examine two popular software libraries used for facial recognition.
Facial Recognition Techniques
In essence, facial recognition is a science of ‘mapping’ faces with the help of biometrics. The resulting unique digital signature is compared to other signatures from a database. The process has four general steps:
Face detection. The first step is to capture a human face in a photo or video. This task is easiest if the person looks straight at the camera. Nonetheless, state-of-art face recognition software can pick a face even if it is angled slightly.
Most face recognition algorithms rely on 2D images, as databases for them are more readily available. However, some of them use 3-D sensors to capture the shape of a face. As a result, they work well despite, for example changing lighting levels.
There are other sophisticated face recognition systems, which also analyze images from thermal cameras. As a result, they can detect distinguishing features of the face even if hats, glasses, etc., obscure it.
The latter can come in especially handy now when the popularity of masks due to the COVID-19 pandemic is skyrocketing. It poses some difficulties for conventional facial recognition software, as the US National Institute of Standards and Technology (NIST) found out. Its researchers tested 89 commercial facial recognition algorithms, digitally applying face masks to photos. Even the best algorithms made from 5% to 50% errors failing to recognize people on “modified” photos.
Face analysis. The main features of the face from the captured image are distinguished and analyzed. Those landmarks are called nodal points. The analysis may include calculating the distance from forehead to chin, the distance between your eyes, etc.
Some systems are sensitive enough to detect and analyze not only facial features but also skin texture. It improves accuracy, providing algorithms with enough data to differentiate even between identical twins.
Digitizing face image. The acquired measurements (biometrics) are converted into a numerical model — a faceprint. Like a thumbprint, a faceprint is unique for each person.
Matching faceprint to a database. The resulting code is compared to other codes from available databases. For instance, the FBI maintains a database known as the Interstate Photo System. Aimed to help law enforcement officials, it contains 36 mln photographs. However, the bureau can also rely on other data sources (particularly information about driver's licenses in 21 US states). As a result, the total amount of photos it has access to is somewhere around 640 mln.
Facial recognition Software Libraries
At present, there are numerous face recognition instruments on the market. This article will provide a brief overview of such popular libraries as Google Cloud Vision API and Amazon Recognition service.
Google Vision provides access to pre-trained machine learning models for images processing. Those algorithms can classify images into millions of categories, read text, detect objects and faces. The latter feature allows spotting multiple faces on images, analyzing key facial features such as eye, mouth, and nose placement. The algorithm assesses people’s emotional states in the picture (like sorrow, happiness, etc.) and whether they are wearing any headwear. You can use both local and remote images.
Google points out that its algorithm does not support actual facial recognition. In other words, with the help of Google Vision API, you can determine whether there is a face on the photo, its key features and whether the person is likely to be happy or sad. It won’t create and match a unique faceprint to a database for you. Still, you can use the provided metadata to do it yourself.
Amazon Rekcognition can analyze both images and video. Its algorithms will estimate the likelihood of presence, location, and orientation of a face. Besides, it can analyze the position of nodal points (for instance, eyes) and determine the person’s age, gender, and emotional state. Moreover, the user can compare a face from one image to a face from another image.
The Use Cases of Face Recognition in Education
Though the full value of facial recognition for education is yet to be harnessed, several ways students and teachers can benefit from new technology have already been discovered. Here are some of them.
Attendance tracking is arguably among the most obvious examples of facial recognition in education. Many institutions spend their resources on this time-consuming practice measuring the students’ engagement levels or trying to meet some government requirements. The first minutes of each class are traditionally devoted to recording attendance. The use of facial recognition will automate and simplify the process.
The accumulated attendance data could help to improve learning environments and experiences. For instance, administrators could modify schedules after determining the most and least popular class dates and times. Schools could try to find the link between scheduling and students’ efficiency.
Accessing Engagement of Students
Teachers can record their students’ reactions during lectures and analyze them using face recognition technology in classrooms. Thus, they can evaluate the emotional dynamics of the class to find the most and the least exciting and engaging parts of lectures. As such data accumulates, schools and universities can analyze them to look for more insights. For instance, they can determine how students learn and which methods of teaching are most effective.
Students can also benefit from the new technology. Such data may help to discover personal strengths and weaknesses for each of them. Thus, more personalized teaching programs may be devised to make learning more pleasant and effective.
“Although the idea of identifying a person through their own personal characteristics is not new, if you think in terms of physical mugshots used by police since the advent of the camera, it is clear that this specific biometric modality is growing in interest. As the demand and need from business owners, law enforcement officials and security professionals intensifies, it is clear the need for proactive and predictive solutions is appropriate and creates a situation where facial recognition technology could possibly fill a needed gap” — says Bill Edwards, Associate Principal of Protective Design and Security services at Thornton Tomasetti.
Using facial recognition technology in education, schools and universities can provide more privacy and physical safety to students and staff. Administrators can compile a database of individuals, which shouldn’t be allowed access to campuses (known offenders, for instance). Then the new technology can be used to scan camera feeds for these people and alert security officers about their location. Some local educational authorities in the US expressed their interest in facial recognition as an additional security measure to prevent gun violence.
The Challenges of Using Facial Recognition Technology in Classrooms
Facial recognition technologies are becoming more efficient and easily available. However, the use of facial recognition software in schools and other educational institutions raises some concerns. They have to be addressed to implement the new tools successfully.
Even the best face recognition technologies can’t be accurate 100% of the time. Varying facial expressions, light, pose, and aging provide technical challenges. Moreover, some algorithms may be biased towards people with a specific appearance.
As the study of the National Institute of Standards and Technology suggests, the performance of facial recognition software varies for different demographic groups. It means that, for instance, African Americans may run a higher risk of being flagged as a security threat. Therefore, algorithms must be configured properly to avoid discrimination.
As data breaches expose the number of information businesses collect about their clients, causing public outrage, privacy laws become stricter. For instance, the EU's General Data Protection Regulation (GDPR), put into effect in 2018, is one of the world’s toughest security and privacy laws.
Capturing and storing information about a person's facial features may be interpreted as gathering personal data. Thus, to use facial recognition, organizations might need explicit consent from the people they will monitor. In the case of schools, for instance, it may mean the consent of parents. Devising a plan on how to implement the face recognition technology in education, don't forget to consider legal implications.
“In any store, venue, or property where you're using facial recognition technology, consent is key. You can receive consent by executing the solution on an opt-in basis and educating visitors on how their image is captured, used, and permanently deleted. This will help you effectively promote privacy and customer reassurance every step of the way,” — says Rob Watts, CEO at Corsight AI.
Some people are against using facial recognition as they are worried about how information about them might be used. However, even if the educational institution has the best intentions, it should also ensure data security. Otherwise sensitive data of its staff and students may be compromised as a result of a cyberattack. In other words, the implementation of facial recognition technology in schools and universities should be closely linked to boosting cybersecurity.
Why Should you Partner with Menklab to Adopt Facial Recognition Software in Edtech?
Menklab has both expertise and experience to help with engineering state-of-the-art digital solutions for the education industry. Our project for early care and preschool service, Nurture, may serve as an example.
Our client wanted to improve the experience of its customers (parents) and staff (teachers). Parents wanted to get the latest information about their children online. At the same time, teachers looked for ways to spend less time doing mundane administrative tasks. Besides, the company needed to minimize errors in daily tasks’ management.
Nurture chose our company to create an end-to-end pre-school communications and management platform for parents and teachers. The project encompassed a web application for school management, iOS apps for teachers and parents, an Android app for parents, and a web payment portal. With the help of our platform, parents can enroll children, manage their accounts, and watch live videos of their offspring. Each child has a Detail page specifying their behavior, mood, meals, naps, curriculum, etc.
Our solution provides physical security of entry and exit points for staff, students, and parents. The location of each student on the premises can be tracked. On top of it, the platform provides emergency communication between the school and parents.
As of January 2021, parents from over 30 countries were using our app. More efficient tuition process, enhanced management of school operations, higher retention rates allowed the client to save more than 5% of topline revenue. The company was able to open two additional schools each year, overfilling its plan.
Create Supreme Software for Education with Menklab
The use of facial recognition technology in education has a large potential. It can assist teachers in collecting valuable feedback and improve their performance. Some routine administrative tasks can be automated, so professors can devote more time to teaching their students. Moreover, students can get better education due to advancements in personalized learning. And finally, facial recognition can make kindergartens, schools, and universities safer.
At the same, the implementation of this new technology is hindered by several challenges. The facial recognition system should be robust enough to cope with different distortions in face images. Also, some people are worried that such systems threaten their privacy. That’s why implementing facial recognition may require complying with some legal procedures. Last but not least is the necessity to provide secure storage for data obtained by new technology.
A reputable, experienced partner can help to overcome those difficulties, especially technical ones. Menklab can be such a partner for you. We’ve already implemented a complex software solution for the education industry. Besides, we love big ideas and new challenges. Let's build a new revolutionary solution for the education industry together!