We've all, or at least most of us, at some point been a proxy for a friend who missed a class and even got away with it. But if there's no roll call how will you be the proxy?
Seven students from Chennai's Great Learning have developed a face-recognition app that takes footage from a camera installed in the room and marks students present or absent and they say it's 99 per cent accurate.
The team — Aruna Kumari Evoori, Vijaya Babu Vemuri, Thangaselvi Arichandrapandian, Karthikeyan G, Anwesh Reddy Paduri, Dhivakar Babu and Jayadev Madhavan — along with their mentor Anand Subramaniam and Narayana Darapaneni, Director, AI and Machine Learning, have also written a paper — Automatic Face Detection and Recognition for Attendance Maintenance — that explain the workings of the app and how Deep Learning helps take any frame and enhance it to match the right person. "We take a photo from the database and augment it to make a three-dimensional image of the person so that the face recognition is possible from any angle. We are making 15-16 separate photos from possible positions that the person can be seen from and then training our model," said Vijaya Babu Vemuri. "The classroom will have a video camera and during each class and we will take screenshots from it at least three to four times to ascertain the attendance. We take the photos multiple times because maybe at that point the person is not in class. We don't want to miss anyone. The reports will come automatically and there is no manual intervention anywhere," he added.
But putting a camera inside the classroom will be a challenge to start with. Many schools and colleges have faced backlash from their students for 'putting them under surveillance. "A lot of schools and colleges now have CCTVs installed. So the surveillance threat is not something we are concerned about but we are definitely taking security very seriously. We need to make sure that the data is handled securely," said Aruna Kumari Evoori.
The project is now in its nascent stage and the team will come up with a cost-effective product, they assured. "To give you an idea, we used a basic camera in our classroom and it could cover a room of 60 students. The software costs are also minimal," said Thangaselvi Arichandrapandian, another member of the team.
