What is AKi?
Every day, hours are wasted taking attendance of students when they enter the class. We identified this problem and decide to find a solution. Aki is a system that leverages the power of Artificial Intelligence and Face Recognition to take attendance of students when they are entering the classroom. We created a web application that receives immediate input from the small cameras in every classroom door, it crosses the information with students' schedules and profiles and creates an accurate report of attendance every single day.
How it works:
When the school decides they want to implement AKi, they will create an account and connect with cameras via WiFi. The school must create the database of students, that is done by making folders with the class number (11th grade), inside that folder there will be folders with the name of each student and inside those folders pictures of the face of the student from different angles. The AI algorithm is already trained to recognize faces so once a student appears on camera it will calculate similarity to all the databases of students and if it the has similarity higher than 60% it assumes that is the student and updates the attendance. Marking that the student is present and the time or arrival. In the website the school administration will have a complete report of every days attendance for every class. Also by searching a student name the school will see a complete report of the student profile and statistics and the same for a whole class. We are going to make available a penalization for being absent and also an option to make an automatic email to a student when absent to ask him the reason. In case that something goes wrong (student face not being identified) there is always a manual input for teachers to correct the database.
How we built it:
1) We brainstormed ideas on how to solve the problem and different features that would be needed to make things easier for the teachers. We came up with interesting things, for instance making every student scan a QR code with their cellphone to inform their presence. After a careful analysis of the options, we decided that Face Recognition was the way to go. 2) We divided the tasks in the team considering everyone's strengths. We used Adobe XD to design the user interface. Then that was passed to someone with experience in web development who started to make the website using HTML CSS JS and Python. At the same time someone else took care of the AI part, at first we started to train a neural network algorithm in Tensorflow ourselves, but then we realized that using OpenCV (open source library for face recognition) was a more efficient and accurate choice. 3) We connected the AI output from the cameras with the website and GUI. Making sure everything worked as planned. 4) Then we looked at the specification of a camera needed for this particular use and used Fusion 360 to make a 3d design of a camera. We want it to be discrete and resistant. 5) Lastly we joined everything together and made a video presentation.
Challenges we ran into:
We definitely encountered challenges along the way but we successfully managed to solve them. When training the algorithm we noticed the Cost (AI metric that should decrease when the algorithm is trained) was not decreasing. Then we changed it using OpenCV. Another challenge was finding a name and logo that was simple, easy to remember, and discrete since it is a system being used at schools. When trying to run the code on our webcams, we found that faces were not being detected and spent hours checking the code when the issue was the camera quality, it needed at least a 1080 camera to be precise. When building the web app we found issues with the responsiveness of it. We were able to make it responsive after spending hours on it.
What we learned and Accomplishments that we're proud of:
We are very proud of what we accomplished in this hackathon and we are excited to continue this amazing project. Since this is not our first hackathon we already had good knowledge on the areas we worked on and organization. Given that we are a small team of 4 people we can move fast and iterate quickly. The basis of our team was respect and communication, everyone was the leader of their task and whenever someone was free, he made himself available to help with the other tasks. Because we believe in what we made, and we are sure that AKi should exist, we were motivated the whole time to keep working in spite of the time. We learned that what may seem impossible at first to do in a short amount of time, is possible if we all work together. We learned a lot about the other members' domains of expertise: Artificial Intelligence, Video editing and animation, Web development, Design and ideation.
What is next for AKi:
We are very proud of what we accomplished in such a short amount of time and we did it happily because we actually think AKi should exist and we want to be the ones making it a reality. The first thing we would like to do is finish the first MVP (Minimum Viable Product) and test it in an actual school environment. This is a tool we are making for teachers so their input is absolutely important, hence we would like them to try the product and give honest feedback on what changes should be made. We would also like to buy a few cameras with the specifications we need and test them in the real setting of use. Making a product is an iterative process and we are eager to start it. Once we have something useful we could probably ask our school for permission to place the system in a few classrooms to keep testing it.
Log in or sign up for Devpost to join the conversation.