2.1 Utilizing face recognition for tracking quarantine evaders during COVID-19 
Recognizing faces is one of the first developed human instincts. Small babies need to recognize the faces of their parents from other faces so they can connect to them. In this context, babies start staring and seeing their parents’ faces (for instance, during breastfeeding) until they memorize them. They can then start seeing the difference between their parents’ faces and unfamiliar faces. Similarly, computers mimic human behavior and intelligence and, instead of using human eyes, they use sensors (cameras) to identify faces. Facial recognition is based on the geometric characteristics of the face (the shape of eyes, nose, mouth and the geometric relationship between them (the distance between them).
| Geometric (/ˌdʒiəˈmɛtrɪk/) Characteristics (/ˌkærɪktəˈrɪstɪks/) |
|---|
| They are visual characteristics that made up of shapes such as squares, triangles, or rectangles. |
Facial recognition involves, as shown in Figure 5, the following four steps.
Face Detection: Locate one or more faces in the image and mark with a bounding box.
Face Alignment: Normalize the face to be consistent with the database, such as geometry and photometrics.
Feature Extraction: Select features from the face that can be used for the recognition task.
Face Recognition: Compare the face to one or more known faces in a prepared database.
| Facial (/ˈfeɪʃəl/) Recognition (/ˌrɛk əgˈnɪʃən/) |
|---|
| Facial recognition is a way of recognizing a human face through technology. A facial recognition system uses biometrics to map facial features from a photograph or video. It compares the information with a database of known faces to find a match. |
Story 2: Using robots for face mask detection during COVID-19
“Your temperature is normal. Please wear a mask”, say AIMBOT. AIMBOT is an intelligent patrol robot that reports and reminds people to wear a mask, as shown in Figure 6. When multiple people enter together, it can also conduct multiple face detections and mask identifications. Then it can remind everyone to keep practicing social distancing in order to reduce the risk of infection. How can AIMBOT do this? Even humans cannot measure their body temperature with only their eyes. It uses an infrared sensor in addition to its visible light binocular camera (like human eyes) to identify the visitor's mask on their faces and body temperature.
| Infrared (/ˌɪnfrəˈrɛd/) |
|---|
| It is a type of light that cannot be seen with the naked eye, but can be felt in the form of heat. |
Story 3: Using face recognition for tracking quarantine evaders during COVID-19
The COVID-19 virus is very tricky. After infecting people, it has an incubation period of about 14 days in the body. An “incubation period” means the amount of time it takes before a person starts showing symptoms, like a fever, or coughing. An incubation period of 14 days means a person can have the virus for two whole weeks before they realize it. You may be worried and asking yourself, "How can I tell if I'm surrounded by infected people? Especially if they do not have any symptoms? How many people are in close contact with them? " Bingo! People's mobile trajectory data can tell us! The trajectory data can tell you where the patient has been and whether you've been there. For example, the Beijing JianKang APP uses the camera to detect the person’s face. It then shows the current status of that person. As shown in Figure 7, the green word indicates that the health of that person is good, while the yellow word indicates that the person should be home quarantined. Many supermarkets, parks and other public places in China used this APP to allow only those who are healthy to enter these places to protect more people.
| Incubation (/ˌɪnkyəˈbeɪʃən, ˌɪŋ-/) Period (/ˈpɪəriəd/) |
|---|
| The period of time between harmful bacteria or viruses entering a person's or animal's body, or entering a plant, and the symptoms of a disease appearing. |
| Mobile (/ˈmoʊbəl, -baɪl/) Trajectory (/trəˈdʒɛktəri/) Data (/ˈdeɪ tə, ˈdætə, ˈdɑtə/) |
|---|
| Data that show the curved path of somebody or something moving from one place to another. |
The following challenges can make it difficult for computers to recognize a face.
- Pose variations: It is sometimes difficult for humans to recognize others, for instance, their friends when they are looking at them from different angles. Similarly, head movements, which can be described by the egocentric rotation angles, i.e. pitch, roll and yaw, or camera changing point of views could lead to major changes in facial appearance and/or shape and generate individual variations (differences), making automated facial recognition across pose a difficult task for computers.
| Variation (/ˌvɛəriˈeɪʃən/) |
|---|
| A change in the amount or level of something. |
- Facial expression changes: Even more differences could be caused by a person making different facial expressions. A person might change their face because they are in different emotional states (happy, angry, confused, etc.) like the ones in Figure 8. So, efficiently and automatically recognizing the different facial expressions is important for both the evaluation of emotional states and the automated face recognition.
Aging of the face: Another reason of face appearance's changes could be caused by the aging of the human face, and could impact on the entire process if the time between each image capture is significant.
Varying illumination conditions: It is sometimes difficult for humans to identify others when it is a bit dark (for instance at night). Similarly, large variations of light and dark could degrade the performance of recognizing a given face. Indeed, for low levels of lighting of the background or foreground, face detection and recognition are much harder to perform, since shadows could appear on the face and facial patterns could look the same, as shown in Figure 9.
