How we made an AI Attendance System for better Team Performance

Our team at Hexmos had problems with absenteeism. When the team involved aren’t consistent, it leads to chaos in managing projects and lower team performance.

This is a common problem among teams and costs billions in losses among employers across the world. So we came up with an attendance system for the team.

Planning the attendance system

To design a proper attendance system, we had to refer to the existing implementations. We can see Attendance done via fingerprint recognition, face recognition or just showing up when the attendance is taken.

Our Team works remotely, so we planned to implement a face recognition attendance system in our Hexmos Feedback Mobile App.

The idea is to scan our face using our mobile app and mark attendance on that particular day. This reduces the efforts from the manager's side, and keeps things more automated.

Building the attendance system

We had a primary problem at hand to implement the attendance system, which is to figure out a way to implement facial recognition.

To achieve this I used a Python library called face_recognition

face_recognition uses deep learning for performing the facial recognition tasks.

Using the face_recognition library we can obtain the following data about the face

• Face encodings

• Face Locations

• Face Landmarks

Face encodings are facial features expressed in numbers using deep learning models.

Face locations are the regions within an image where faces are detected.

Face landmarks are key features on a face, such as the eyes, nose, mouth, and chin.

When a user registers his/her face, we will insert the above 3 data into the Database. This data will be unique for everyone and hence will be beneficial for facial recognition

When the user tries to face recognize, The inputted image's face encodings will be compared with that of the one stored in the Database.

If the match is confirmed, it indicates that the faces are the same, So we can mark the attendance for that particular person.

Implementing the attendance system

For the attendance implementation, we had to do 2 steps

• Register the user's face in the Hexmos Feedback app

• Do facial recognition every day to mark attendance.

We also added a location component, to keep track of the location where the attendance is being taken.

Conclusion

This is just the start of our AI-based attendance system, there are many more things that can be developed to reduce the pain and chaos involved in managing absenteeism, such as more detailed statistics, notifications of checking in, leave management and many more.